No Arabic abstract
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 (~3700-7000 Angstrom) 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., [OII], Hbeta, [OIII], 5008, Halpha, [NII], and [SII]) and stellar continuum and absorption features (e.g., Balmer lines, Ca-II H and K, Mgb, 4000 Angstrom 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 Msol) and star formation rate (~10^0-10^3 Msol/yr). 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 use extensive spectroscopy from the MOSFIRE Deep Evolution Field (MOSDEF) survey to investigate the relationships between rest-frame optical emission line equivalent widths ($W$) and a number of galaxy and ISM characteristics for a sample of $1134$ star-forming galaxies at redshifts $1.4lesssim zlesssim 3.8$. We examine how the equivalent widths of [OII]$lambdalambda 3727, 3730$, H$beta$, [OIII]$lambdalambda 4960, 5008$, [OIII]$+$H$beta$, H$alpha$, and H$alpha$+[NII]$lambdalambda 6550, 6585$, depend on stellar mass, UV slope, age, star-formation rate (SFR) and specific SFR (sSFR), ionization parameter and excitation conditions (O32 and [OIII]/H$beta$), gas-phase metallicity, and ionizing photon production efficiency ($xi_{rm ion}$). The trend of increasing $W$ with decreasing stellar mass is strongest for [OIII] (and [OIII]+H$beta$). More generally, the equivalent widths of all the lines increase with redshift at a fixed stellar mass or fixed gas-phase metallicity, suggesting that high equivalent width galaxies are common at high redshift. This redshift evolution in equivalent widths can be explained by the increase in SFR and decrease in metallicity with redshift at a fixed stellar mass. Consequently, the dependence of $W$ on sSFR is largely invariant with redshift, particularly when examined for galaxies of a given metallicity. Our results show that high equivalent width galaxies, specifically those with high $W({rm [OIII]})$, have low stellar masses, blue UV slopes, young ages, high sSFRs, ISM line ratios indicative of high ionization parameters, high $xi_{rm ion}$, and low metallicities. As these characteristics are often attributed to galaxies with high ionizing escape fractions, galaxies with high $W$ are likely candidates for the population that dominates cosmic reionization.
We have used FMOS on Subaru to obtain near-infrared spectroscopy of 123 far-infrared selected galaxies in COSMOS and obtain the key rest-frame optical emission lines. This is the largest sample of infrared galaxies with near-infrared spectroscopy at these redshifts. The far-infrared selection results in a sample of galaxies that are massive systems that span a range of metallicities in comparison with previous optically selected surveys, and thus has a higher AGN fraction and better samples the AGN branch. We establish the presence of AGN and starbursts in this sample of (U)LIRGs selected as Herschel-PACS and Spitzer-MIPS detections in two redshift bins (z~0.7 and z~1.5) and test the redshift dependence of diagnostics used to separate AGN from star-formation dominated galaxies. In addition, we construct a low redshift (z~0.1) comparison sample of infrared selected galaxies and find that the evolution from z~1.5 to today is consistent with an evolving AGN selection line and a range of ISM conditions and metallicities from the models of Kewley et al. (2013b). We find that a large fraction of (U)LIRGs are BPT-selected AGN using their new, redshift-dependent classification line. We compare the position of known X-ray detected AGN (67 in total) with the BPT selection and find that the new classification line accurately selects most of these objects (> 70%). Furthermore, we identify 35 new (likely obscured) AGN not selected as such by their X-ray emission. Our results have direct implications for AGN selection at higher redshift with either current (MOSFIRE, KMOS) or future (PFS, MOONS) spectroscopic efforts with near-infrared spectral coverage.
We present K-band spectra of rest-frame optical emission lines for 24 star-forming galaxies at z~3.2-3.7 using MOSFIRE on the Keck 1 telescope. Strong rest-frame optical [O III] and Hbeta emission lines were detected in 18 LBGs. The median flux ratio of [O III]5007 to Hbeta is 5.1+/-0.5, a factor of 5-10x higher than in local galaxies with similar stellar masses. The observed Hbeta luminosities are in good agreement with expectations from the estimated star-formation rates, and none of our sources are detected in deep X-ray stacks, ruling out significant contamination by active galactic nuclei. Combining our sample with a variety of LBGs from the literature, including 49 galaxies selected in a very similar manner, we find a high median ratio of [OIII]/Hbeta = 4.8+0.8-1.7. This high ratio seems to be an ubiquitous feature of z~3-4 LBGs, very different from typical local star-forming galaxies at similar stellar masses. The only comparable systems at z~0 are those with similarly high specific star-formation rates, though ~5x lower stellar masses. High specific star-formation rates either result in a much higher ionization parameter or other unusual conditions for the interstellar medium, which result in a much higher [OIII]/Hbeta line ratio. This implies a strong relation between a global property of a galaxy, the specific star-formation rate, and the local conditions of ISM in star-forming regions.
Euclid, WFIRST, and HETDEX will make emission-line selected galaxies the largest observed constituent in the $z > 1$ universe. However, we only have a limited understanding of the physical properties of galaxies selected via their Ly$alpha$ or rest-frame optical emission lines. To begin addressing this problem, we present the basic properties of $sim 2,000$ AEGIS, COSMOS, GOODS-N, GOODS-S, and UDS galaxies identified in the redshift range $1.90 < z < 2.35$ via their [O II], H$beta$, and [O III] emission lines. For these $z sim 2$ galaxies, [O III] is generally much brighter than [O II] and H$beta$, with typical rest-frame equivalent widths of several hundred Angstroms. Moreover, these strong emission-line systems span an extremely wide range of stellar mass ($sim 3$ dex), star-formation rate ($sim 2$ dex), and [O III] luminosity ($sim 2$ dex). Comparing the distributions of these properties to those of continuum selected galaxies, we find that emission-line galaxies have systematically lower stellar masses and lower optical/UV dust attenuations. These measurements lay the groundwork for an extensive comparison between these rest-frame optical emission-line galaxies and Ly$alpha$ emitters identified in the HETDEX survey.
The complex structure of gas, metals, and dust in the interstellar and circumgalactic medium (ISM and CGM, respectively) in star-forming galaxies can be probed by Ly$alpha$ emission and absorption, low-ionization interstellar (LIS) metal absorption, and dust reddening E(B-V). We present a statistical analysis of the mutual correlations among Ly$alpha$ equivalent width (EW$_{Lyalpha}$), LIS equivalent width (EW$_{LIS}$), and E(B-V) in a sample of 157 star-forming galaxies at $zsim2.3$. With measurements obtained from individual, deep rest-UV spectra and spectral-energy distribution (SED) modeling, we find that the tightest correlation exists between EW$_{LIS}$ and E(B-V), although correlations among all three parameters are statistically significant. These results signal a direct connection between dust and metal-enriched HI gas, and that they are likely co-spatial. By comparing our results with the predictions of different ISM/CGM models, we favor a dusty ISM/CGM model where dust resides in HI gas clumps and Ly$alpha$ photons escape through the low HI covering fraction/column density intra-clump medium. Finally, we investigate the factors that potentially contribute to the intrinsic scatter in the correlations studied in this work, including metallicity, outflow kinematics, Ly$alpha$ production efficiency, and slit loss. Specifically, we find evidence that scatter in the relationship between EW$_{Lyalpha}$ and E(B-V) reflects the variation in metal-to-HI covering fraction ratio as a function of metallicity, and the effects of outflows on the porosity of the ISM/CGM. Future simulations incorporating star-formation feedback and the radiative transfer of Ly$alpha$ photons will provide key constraints on the spatial distributions of neutral hydrogen gas and dust in the ISM/CGM structure.