In this paper we present the MOSFIRE Deep Evolution Field (MOSDEF) survey. The MOSDEF survey aims to obtain moderate-resolution (R = 3000 − 3650) rest-frame optical spectra (∼ 3700 − 7000Å) for ∼1500 galaxies at 1.37 ≤ z ≤ 3.80 in three well-studied CANDELS fields: AEGIS, COSMOS, and GOODS-N. Targets are selected in three redshift intervals: 1.37 ≤ z ≤ 1.70, 2.09 ≤ z ≤ 2.61, and 2.95 ≤ z ≤ 3.80, down to fixed H AB (F160W) magnitudes of 24.0, 24.5 and 25.0, respectively, using the photometric and spectroscopic catalogs from the 3D-HST survey. We target both strong nebular emission lines (e.g.,, 6585, and [S ii] λλ6718, 6733) and stellar continuum and absorption features (e.g., Balmer lines, Ca-ii H and K, Mgb, 4000Å break). Here we present an overview of our survey, the observational strategy, the data reduction and analysis, and the sample characteristics based on spectra obtained during the first 24 nights. To date, we have completed 21 masks, obtaining spectra for 591 galaxies. For ∼80% of the targets we derive a robust redshift from either emission or absorption lines. In addition, we confirm 55 additional galaxies, which were serendipitously detected. The MOSDEF galaxy sample includes unobscured star-forming, dusty star-forming, and quiescent galaxies and spans a wide range in stellar mass (∼ 10 9 − 10 11.5 M ⊙ ) and star formation rate (∼ 10 0 − 10 3 M ⊙ yr −1 ). The spectroscopically confirmed sample is roughly representative of an H-band limited galaxy sample at these redshifts. With its large sample size, broad diversity in galaxy properties, and wealth of available ancillary data, MOSDEF will transform our understanding of the stellar, gaseous, metal, dust, and black hole content of galaxies during the time when the universe was most active.

We investigate the evolution of galaxy gas-phase metallicity (O/H) over the range z = 0 − 3.3 using samples of ∼ 300 galaxies at z ∼ 2.3 and ∼ 150 galaxies at z ∼ 3.3 from the MOSDEF survey. This analysis crucially utilizes different metallicity calibrations at z ∼ 0 and z > 1 to account for evolving ISM conditions. We find significant correlations between O/H and stellar mass (M * ) at z ∼ 2.3 and z ∼ 3.3. The low-mass power law slope of the mass-metallicity relation is remarkably invariant over z = 0 − 3.3, such that O/H∝M 0.30 * at all redshifts in this range. At fixed M * , O/H decreases with increasing redshift as dlog(O/H)/dz = −0.11 ± 0.02. We find no evidence that the fundamental metallicity relation between M * , O/H, and star-formation rate (SFR) evolves out to z ∼ 3.3. We employ analytic chemical evolution models to place constraints on the mass and metal loading factors of galactic outflows. The efficiency of metal removal increases toward lower M * at fixed redshift, and toward higher redshift at fixed M * . These models suggest that the slope of the mass-metallicity relation is primarily set by the scaling of the outflow metal loading factor with M * , not by the change in gas fraction as a function of M * . The evolution toward lower O/H at fixed M * with increasing redshift is driven by both higher gas fraction (leading to stronger dilution of ISM metals) and higher metal removal efficiency. These results suggest that the processes governing the smooth baryonic growth of galaxies via gas flows and star formation hold in the same form over at least the past 12 Gyr.

We present detections of [O iii]λ4363 and direct-method metallicities for star-forming galaxies at z = 1.7 − 3.6. We combine new measurements from the MOSFIRE Deep Evolution Field (MOSDEF) survey with literature sources to construct a sample of 18 galaxies with direct-method metallicities at z > 1, spanning 7.5 < 12+log(O/H) < 8.2 and log(M * /M ) = 7 − 10. We find that strong-line calibrations based on local analogs of high-redshift galaxies reliably reproduce the metallicity of the z > 1 sample on average. We construct the first mass-metallicity relation at z > 1 based purely on direct-method O/H, finding a slope that is consistent with strong-line results. Directmethod O/H evolves by 0.1 dex at fixed M * and SFR from z ∼ 0 − 2.2. We employ photoionization models to constrain the ionization parameter and ionizing spectrum in the high-redshift sample. Stellar models with super-solar O/Fe and binary evolution of massive stars are required to reproduce the observed strong-line ratios. We find that the z > 1 sample falls on the z ∼ 0 relation between ionization parameter and O/H, suggesting no evolution of this relation from z ∼ 0 to z ∼ 2. These results suggest that the offset of the strong-line ratios of this sample from local excitation sequences is driven primarily by a harder ionizing spectrum at fixed nebular metallicity compared to what is typical at z ∼ 0, naturally explained by super-solar O/Fe values at high redshift caused by rapid formation timescales. Given the extreme nature of our z > 1 sample, the implications for representative z ∼ 2 galaxy samples at ∼ 10 10 M are unclear, but similarities to z > 6 galaxies suggest that these conclusions can be extended to galaxies in the epoch of reionization.

We investigate the nature of the relation among stellar mass, star-formation rate, and gas-phase metallicity (the M * -SFR-Z relation) at high redshifts using a sample of 260 star-forming galaxies at z ∼ 2.3 from the MOSDEF survey. We present an analysis of the high-redshift M * -SFR-Z relation based on several emission-line ratios for the first time. We show that a M * -SFR-Z relation clearly exists at z ∼ 2.3. The strength of this relation is similar to predictions from cosmological hydrodynamical simulations. By performing a direct comparison of stacks of z ∼ 0 and z ∼ 2.3 galaxies, we find that z ∼ 2.3 galaxies have ∼ 0.1 dex lower metallicity at fixed M * and SFR. In the context of chemical evolution models, this evolution of the M * -SFR-Z relation suggests an increase with redshift of the mass-loading factor at fixed M * , as well as a decrease in the metallicity of infalling gas that is likely due to a lower importance of gas recycling relative to accretion from the intergalactic medium at high redshifts. Performing this analysis simultaneously with multiple metallicity-sensitive line ratios allows us to rule out the evolution in physical conditions (e.g., N/O ratio, ionization parameter, and hardness of the ionizing spectrum) at fixed metallicity as the source of the observed trends with redshift and with SFR at fixed M * at z ∼ 2.3. While this study highlights the promise of performing high-order tests of chemical evolution models at high redshifts, detailed quantitative comparisons ultimately await a full understanding of the evolution of metallicity calibrations with redshift.

We present results on the star formation rate (SFR) versus stellar mass (M * ) relation (i.e., the "main sequence") among star-forming galaxies at 1.37 z 2.61 using the MOSFIRE Deep Evolution Field (MOSDEF) survey. Based on a sample of 261 galaxies with Hα and Hβ spectroscopy, we have estimated robust dust-corrected instantaneous SFRs over a large range in M * (∼10 9.5 -10 11.5 M e ). We find a correlation between log(SFR(Hα)) and log(M * ) with a slope of 0.65 ± 0.08 (0.58 ± 0.10) at 1.4 < z < 2.6 (2.1 < z < 2.6). We find that different assumptions for the dust correction, such as using the colorexcess of the stellar continuum to correct the nebular lines, sample selection biases against red star-forming galaxies, and not accounting for Balmer absorption, can yield steeper slopes of the log(SFR)-log(M * ) relation. Our sample is immune from these biases as it is rest-frame optically selected, Hα and Hβ are corrected for Balmer absorption, and the Hα luminosity is dustcorrected using the nebular colorexcess computed from the Balmer decrement. The scatter of the log(SFR(Hα))-log(M * ) relation, after accounting for the measurement uncertainties, is 0.31 dex at 2.1 < z < 2.6, which is 0.05 dex larger than the scatter in log(SFR(UV))-log(M * ). Based on comparisons to a simulated SFR-M * relation with some intrinsic scatter, we argue that in the absence of direct measurements of galaxy-to-galaxy variations in the attenuation/ extinction curves and the initial mass function, one cannot use the difference in the scatter of the SFR(Hα)-and SFR(UV)-M * relations to constrain the stochasticity of star formation in high-redshift galaxies.