Dawn K. Erb

The Mass-Metallicity Relation at z≳2*

We use a sample of 87 rest-frame UV-selected star-forming galaxies with mean spectroscopic redshift z = 2.26 ± 0.17 to study the correlation between metallicity and stellar mass at high redshift. Using stellar masses determined from SED fitting to observed 0.3-8 μm photometry, we divide the sample into six bins in stellar mass and construct six composite Hα + [N ] spectra from all of the objects in each bin. We estimate the mean oxygen abundance in each bin from the [N II]/Hα ratio and find a monotonic increase in metallicity with increasing stellar mass, from 12 + log(O/H) < 8.2 for galaxies with M = 2.7 × 109 M☉ to 12 + log(O/H) = 8.6 for galaxies with M = 1.0 × 1011 M☉. We use the empirical relation between SFR density and gas density to estimate the gas fractions of the galaxies, finding an increase in gas fraction with decreasing stellar mass. These gas fractions, combined with the observed metallicities, allow the estimation of the effective yield yeff as a function of stellar mass; in constrast to observations in the local universe, which show a decrease in yeff with decreasing baryonic mass, we find a slight increase. Such a variation of metallicity with gas fraction is best fitted by a model with supersolar yield and an outflow rate ~4 times higher than the SFR. We conclude that the mass-metallicity relation at high redshift is driven by the increase in metallicity as the gas fraction decreases through star formation and is likely modulated by metal loss from strong outflows in galaxies of all masses.

The Stellar, Gas, and Dynamical Masses of Star-forming Galaxies at z ~ 2

We present analysis of the near-infrared spectra of 114 rest-frame UV-selected star-forming galaxies at z ~ 2. By combining the Hα spectra with photometric measurements from observed 0.3-8 μm, we assess the relationships among kinematics, dynamical masses, inferred gas fractions, and stellar masses and ages. The Hα line widths give a mean dynamical mass M_(dyn) = (6.9 ± 0.6) × 10^(10) M_⊙ within a typical radius of ~6 kpc, after excluding AGNs. The average dynamical mass is ~2 times larger than the average stellar mass, and the two agree to within a factor of several for most objects. However, ~15% of the sample has M_(dyn)≫ M. These objects are best fit by young stellar populations and tend to have high Hα equivalent widths, W_(Hα) ≳ 200 A, suggesting that they are young starbursts with large gas masses. Rest-frame optical luminosity and velocity dispersion are correlated with 4 σ significance. Using the local empirical correlation between star formation rate per unit area and gas surface density, we estimate the mass of the gas associated with star formation and find a mean gas fraction of ~50% and a strong decrease in gas fraction with increasing stellar mass. The masses of gas and stars combined are considerably better correlated with the dynamical masses than are the stellar masses alone, and agree to within a factor of 3 for 85% of the sample. The combination of kinematic measurements, estimates of gas masses, and stellar population properties suggest that the factor of ~500 range in stellar mass across the sample cannot be fully explained by intrinsic differences in the total masses of the galaxies, which vary by a factor of ~40; the remaining variation is due to the evolution of the stellar population and the conversion of gas into stars.

Hα Observations of a Large Sample of Galaxies at z ~ 2: Implications for Star Formation in High-Redshift Galaxies*

Using Hα spectra of 114 rest-frame UV-selected galaxies at z ~ 2, we compare inferred star formation rates (SFRs) with those determined from the UV continuum luminosity. After correcting for extinction using standard techniques based on the UV continuum slope, we find excellent agreement between the indicators, with = 31 M_☉ yr^(-1) and = 29 M_☉ yr^(-1). The agreement between the indicators suggests that the UV luminosity is attenuated by a typical factor of ~4.5 (ranging from no attenuation to a factor of ~100 for the most obscured object in the sample), in good agreement with estimates of obscuration from X-ray, radio, and mid-IR data. The Hα luminosity is attenuated by a factor of ~1.7 on average, and the maximum Hα attenuation is a factor of ~5. In agreement with X-ray and mid-IR studies, we find that the SFR increases with increasing stellar mass and at brighter K magnitudes to ~ 60 M_☉ yr^(-1) for galaxies with K_s 10^(11) M_☉) have had higher SFRs in the past.

Multiwavelength Constraints on the Cosmic Star Formation History from Spectroscopy: The Rest-Frame Ultraviolet, Hα, and Infrared Luminosity Functions at Redshifts 1.9 ≲ z ≲ 3.4*

We use a sample of rest-frame UV-selected and spectroscopically observed galaxies at redshifts 1.9 ≤ z < 3.4, combined with ground-based spectroscopic Hα and Spitzer MIPS 24 μm data, to derive the most robust measurements of the rest-frame UV, Hα, and infrared (IR) luminosity functions (LFs) at these redshifts. Our sample is by far the largest of its kind, with over 2000 spectroscopic redshifts in the range 1.9 ≤ z < 3.4 and ~15,000 photometric candidates in 29 independent fields covering a total area of almost a square degree. Our method for computing the LFs takes into account a number of systematic effects, including photometric scatter, Lyα line perturbations to the observed optical colors of galaxies, and contaminants. Taking into account the latter, we find no evidence for an excess of UV-bright galaxies over what was inferred in early z ~ 3 LBG studies. The UV LF appears to undergo little evolution between z ~ 4 and z ~ 2. Corrected for extinction, the UV luminosity density (LD) at z ~ 2 is at least as large as the value at z ~ 3 and a factor of ~9 larger than the value at z ~ 6, primarily reflecting an increase in the number density of bright galaxies between z ~ 6 and z ~ 2. Our analysis yields the first constraints anchored by extensive spectroscopy on the infrared and bolometric LFs for faint and moderately luminous (L_(bol) ≾ 10^(12) L⊙) galaxies. Adding the IR to the emergent UV luminosity, incorporating independent measurements of the LD from ULIRGs, and assuming realistic dust attenuation values for UV-faint galaxies, indicates that galaxies with L_(bol) < 10^(12) L⊙ account for ≈80% of the bolometric LD and SFRD at z ~ 2–3. This suggests that previous estimates of the faint end of the Lbol LF may have underestimated the steepness of the faint-end slope at L_(bol) < 10^(12) L⊙. Our multiwavelength constraints on the global SFRD indicate that approximately one-third of the present-day stellar mass density was formed in subultraluminous galaxies between redshifts z = 1.9–3.4.

Ultraviolet to Mid-Infrared Observations of Star-forming Galaxies at z ~ 2: Stellar Masses and Stellar Populations*

We present the broadband UV through mid-infrared spectral energy distributions (SEDs) of a sample of 72 spectroscopically confirmed star-forming galaxies at z = 2.30 ? 0.3. Located in a 72 arcmin2 field centered on the bright background QSO, HS 1700+643, these galaxies were preselected to lie at z ~ 2 solely on the basis of their rest-frame UV colors and luminosities and should be representative of UV-selected samples at high redshift. In addition to deep ground-based photometry spanning from 0.35 to 2.15 ?m, we make use of Spitzer IRAC data, which probe the rest-frame near-IR at z ~ 2. The range of stellar populations present in the sample is investigated with simple, single-component stellar population synthesis models. The inability to constrain the form of the star formation history limits our ability to determine the parameters of extinction, age, and star formation rate without using external multiwavelength information. Emphasizing stellar mass estimates, which are much less affected by these uncertainties, we find log M*/M? = 10.32 ? 0.51 for the sample. The addition of Spitzer IRAC data as a long-wavelength baseline reduces stellar mass uncertainties by a factor of 1.5-2 relative to estimates based on optical-Ks photometry alone. However, the total stellar mass estimated for the sample is remarkably insensitive to the inclusion of IRAC data. We find correlations between stellar mass and rest-frame R band (observed Ks) and rest-frame 1.4 ?m (observed 4.5 ?m) luminosities, although with significant scatter. Even at rest-frame 1.4 ?m, the mass-to-light ratio varies by a factor of 15 indicating that even the rest-frame near-IR, when taken alone, is a poor indicator of stellar mass in star-forming galaxies at z ~ 2. Allowing for the possibility of episodic star formation, we find that typical galaxies in our sample could contain up to 3 times more stellar mass in an old underlying burst than what was inferred from single-component modeling. In contrast, mass estimates for the most massive galaxies in the sample (M* > 1011 M?) are fairly insensitive to the method used to model the stellar population. Galaxies in this massive tail, which are also the oldest objects in the sample, could plausibly evolve into the passive galaxies discovered at z ~ 1.7 with near-IR selection techniques. In the general framework of hierarchical galaxy formation and mergers, which implies episodic star formation histories, galaxies at high redshift may pass in and out of UV-selected and near-IR color-selected samples as they evolve from phases of active star formation to quiescence and back again.

THE KILOPARSEC-SCALE KINEMATICS OF HIGH-REDSHIFT STAR-FORMING GALAXIES

We present the results of a spectroscopic survey of the kinematic structure of star-forming galaxies at redshift z ∼ 2–3 using Keck/OSIRIS integral field spectroscopy. Our sample is comprised of 12 galaxies between redshifts z ∼ 2.0 and 2.5 and one galaxy at z ∼ 3.3 which are well detected in either Hα or [O iii] emission. These galaxies are generally representative of the mean stellar mass of star-forming galaxies at similar redshifts, although they tend to have star formation rate surface densities slightly higher than the mean. These observations were obtained in conjunction with the Keck laser guide star adaptive optics system, with a typical angular resolution after spatial smoothing ∼0. 15 (approximately 1 kpc at the redshift of the target sample). At most five of these 13 galaxies have spatially resolved velocity gradients consistent with rotation while the remaining galaxies have relatively featureless or irregular velocity fields. All of our galaxies show local velocity dispersions ∼60–100 km s −1 , suggesting that (particularly for those galaxies with featureless velocity fields) rotation about a preferred axis may not be the dominant mechanism of physical support. While some galaxies show evidence for major mergers such evidence is unrelated to the kinematics of individual components (one of our strongest merger candidates also exhibits unambiguous rotational structure), refuting a simple bimodal disk/merger classification scheme. We discuss these data in light of complementary surveys and extant UV-IR spectroscopy and photometry, concluding that the dynamical importance of cold gas may be the primary factor governing the observed kinematics of z ∼ 2 galaxies. We conclude by speculating on the importance of mechanisms for accreting low angular momentum gas and the early formation of quasi-spheroidal systems in the young universe.

Star formation and extinction in redshift z~2 galaxies: inferences from spitzer mips observations

We use very deep Spitzer MIPS 24 ?m observations to examine the bolometric luminosities (Lbol) and UV extinction properties of more than 200 spectroscopically identified, optically selected (UnG) z ~ 2 galaxies, supplemented with near-IR-selected (BzK and DRG) and submillimeter galaxies at similar redshifts, in the GOODS-N field. Focusing on redshifts 1.5 1012 L?, with a mean LIR 2 ? 1011 L?. Using 24 ?m observations as an independent probe of dust extinction, we find that, as in the local universe, the obscuration LIR/L1600 is strongly dependent on Lbol and ranges in value from <1 to ~1000 within the sample considered. However, the obscuration is generally ~10 times smaller at a given Lbol at z ~ 2 than at z ~ 0. We show that the values of LIR and obscuration inferred from the UV spectral slope ? generally agree well with the values inferred from L5-8.5 ?m for Lbol < 1012 L?. Using the specific SFRs of galaxies as a proxy for cold gas fraction, we find a wide range in the evolutionary state of galaxies at z ~ 2, from galaxies that have just begun to form stars to those that have already accumulated most of their stellar mass and are about to become, or already are, passively evolving.

The Direct Detection of Lyman-Continuum Emission from Star-forming Galaxies at z~3

We present the results of rest-frame UV spectroscopic observations of a sample of 14 z ~ 3 star-forming galaxies in the SSA 22a field. These spectra are characterized by unprecedented depth in the Lyman continuum region. For the first time, we have detected escaping ionizing radiation from individual galaxies at high redshift, with 2 of the 14 objects showing significant emission below the Lyman limit. We also measured the ratio of emergent flux density at 1500 A to that in the Lyman continuum region, for the individual detections (C49 and D3) and the sample average. If a correction for the average IGM opacity is applied to the spectra of the objects C49 and D3, we find f_(1500)/f_(900,corr,C49) = 4.5 and f_(1500)/f_(900,corr,D3) = 2.9. The average emergent flux density ratio in our sample is = 22, implying an escape fraction ~4.5 times lower than inferred from the composite spectrum from Steidel and coworkers. If this new estimate is representative of LBGs, their contribution to the metagalactic ionizing radiation field is J_ν(900) ~ 2.6 × 10^(-22) ergs s^(-1) cm^(-2) Hz^(-1) sr^(-1), comparable to the contribution of optically selected quasars at the same redshift. The sum of the contributions from galaxies and quasars is consistent with recent estimates of the level of the ionizing background at z ~ 3, inferred from the H I Lyα forest optical depth. There is significant variance among the emergent far-UV spectra in our sample, yet the factors controlling the detection or nondetection of Lyman continuum emission from galaxies are not well determined. Because we do not yet understand the source of this variance, significantly larger samples will be required to obtain robust constraints on the galaxy contribution to the ionizing background at z ~ 3 and beyond.

OPTICAL SELECTION OF STAR-FORMING GALAXIES AT REDSHIFTS 1 < z < 3

Few galaxies have been found between the redshift ranges z 1 probed by magnitude-limited surveys and z 3 probed by Lyman break surveys. Comparison of galaxy samples at lower and higher redshift suggests that large numbers of stars were born and the Hubble sequence began to take shape at the intermediate redshifts 1 < z < 3, but observational challenges have prevented us from observing the process in much detail. We present simple and efficient strategies that can be used to find large numbers of galaxies throughout this important but unexplored redshift range. All the strategies are based on selecting galaxies for spectroscopy on the basis of their colors in ground-based images taken through a small number of optical filters: Gi for redshifts 0.85 < z < 1.15, Gz for 1 < z < 1.5, and UnG for 1.4 < z < 2.1 and 1.9 < z < 2.7. The performance of our strategies is quantified empirically through spectroscopy of more than 2000 galaxies at 1 < z < 3.5. We estimate that more than half of the UV luminosity density at 1 < z < 3 is produced by galaxies that satisfy our color selection criteria. Our methodology is described in detail, allowing readers to devise analogous selection criteria for other optical filter systems.

THE SINS SURVEY: MODELING THE DYNAMICS OF z ∼ 2 GALAXIES AND THE HIGH-z TULLY-FISHER RELATION*

We present the modeling of SINFONI integral field dynamics of 18 star-forming galaxies at z ~ 2 from Hα line emission. The galaxies are selected from the larger sample of the SINS survey, based on the prominence of ordered rotational motions with respect to more complex merger-induced dynamics. The quality of the data allows us to carefully select systems with kinematics dominated by rotation, and to model the gas dynamics across the whole galaxy using suitable exponential disk models. We obtain a good correlation between the dynamical mass and the stellar mass, finding that large gas fractions (M gas ≈ M *) are required to explain the difference between the two quantities. We use the derived stellar mass and maximum rotational velocity V max from the modeling to construct for the first time the stellar mass Tully-Fisher relation at z ~ 2.2. The relation obtained shows a slope similar to what is observed at lower redshift, but we detect an evolution of the zero point. We find that at z ~ 2.2 there is an offset in log(M *) for a given rotational velocity of 0.41 ± 0.11 with respect to the local universe. This result is consistent with the predictions of the latest N-body/hydrodynamical simulations of disk formation and evolution, which invoke gas accretion onto the forming disk in filaments and cooling flows. This scenario is in agreement with other dynamical evidence from SINS, where gas accretion from the halo is required to reproduce the observed properties of a large fraction of the z ~ 2 galaxies. Based on observations obtained at the Very Large Telescope (VLT) of the European Southern Observatory, Paranal, Chile, in the context of guaranteed time programs 073.B-9018, 074.A-9011, 075.A-0466, 076.A-0527, 077.A-0576, 078.A-0600, 078.A-0055, 079.A-0341, 080.A-0330, and 080.A-0635.