We constrain the slope of the star formation rate (log Ψ) to stellar mass (log M ⋆ ) relation down to log(M ⋆ /M ⊙ ) = 8.4 (log(M ⋆ /M ⊙ ) = 9.2) at z = 0.5 (z = 2.5) with a mass-complete sample of 39,106 star-forming galaxies selected from the 3D-HST photometric catalogs, using deep photometry in the CANDELS fields. For the first time, we find that the slope is dependent on stellar mass, such that it is steeper at low masses (log Ψ ∝ log M ⋆ ) than at high masses (log Ψ ∝ (0.3 − 0.6) log M ⋆ ). These steeper low mass slopes are found for three different star formation indicators: the combination of the ultraviolet (UV) and infrared (IR), calibrated from a stacking analysis of Spitzer/MIPS 24µm imaging; β-corrected UV SFRs; and Hα SFRs. The normalization of the sequence evolves differently in distinct mass regimes as well: for galaxies less massive than log(M ⋆ /M ⊙ ) < 10 the specific SFR (Ψ/M ⋆ ) is observed to be roughly self-similar with Ψ/M ⋆ ∝ (1 + z) 1.9 , whereas more massive galaxies show a stronger evolution with Ψ/M ⋆ ∝ (1 + z) 2.2−3.5 for log(M ⋆ /M ⊙ ) = 10.2 − 11.2. The fact that we find a steep slope of the star formation sequence for the lower mass galaxies will help reconcile theoretical galaxy formation models with the observations.

We report Hubble Space Telescope/Cosmic Origins Spectrograph observations of the Lyα emission and interstellar absorption lines in a sample of ten star-forming galaxies at z ∼ 0.2. Selected on the basis of high equivalent width optical emission lines, the sample, dubbed "Green Peas," make some of the best analogs for young galaxies in an early Universe. We detect Lyα emission in all ten galaxies, and 9/10 show double-peaked line profiles suggestive of low H I column density. We measure Lyα/Hα flux ratios of 0.5-5.6, implying that 5% to 60% of Lyα photons escape the galaxies. These data confirm previous findings that low-ionization metal absorption (LIS) lines are weaker when Lyα escape fraction and equivalent width are higher. However, contrary to previously favored interpretations of this trend, increased Lyα output cannot be the result of a varying H I covering: the Lyman absorption lines (Lyβ and higher) show a covering fraction near unity for gas with N H I 10 16 cm −2 . Moreover, we detect no correlation between Lyα escape and the outflow velocity of the LIS lines, suggesting that kinematic effects do not explain the range of Lyα/Hα flux ratios in these galaxies. In contrast, we detect a strong anti-correlation between the Lyα escape fraction and the velocity separation of the Lyα emission peaks, driven primarily by the velocity of the blue peak. As this velocity separation is sensitive to H I column density, we conclude that Lyα escape in these Green Peas is likely regulated by the H I column density rather than outflow velocity or H I covering fraction. * Based on observations made with the NASA/ESA Hubble Space Telescope, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 5-26555. These observations are associated with program 12928 and 11727.

We study large-scale outflows in a sample of 96 star-forming galaxies at 1 z 2, using near-UV spectroscopy of Fe II and Mg II absorption and emission. The average blueshift of the Fe II interstellar absorption lines with respect to the systemic velocity is −85 ± 10 km s −1 at z ∼ 1.5, with standard deviation 87 km s −1 ; this is a decrease of a factor of two from the average blueshift measured for far-UV interstellar absorption lines in similarly selected galaxies at z ∼ 2. The profiles of the Mg II λλ2796, 2803 lines show much more variety than the Fe II profiles, which are always seen in absorption; Mg II ranges from strong emission to pure absorption, with emission more common in galaxies with blue UV slopes and at lower stellar masses. Outflow velocities, as traced by the centroids and maximum extent of the absorption lines, increase with increasing stellar mass with 2-3σ significance, in agreement with previous results. We study fine structure emission from Fe II*, finding several lines of evidence in support of the model in which this emission is generated by the re-emission of continuum photons absorbed in the Fe II resonance transitions in outflowing gas. In contrast, photoionization models indicate that Mg II emission arises from the resonant scattering of photons produced in H II regions, accounting for the differing profiles of the Mg II and Fe II lines. A comparison of the strengths of the Fe II absorption and Fe II* emission lines indicates that massive galaxies have more extended outflows and/or greater extinction, while two-dimensional composite spectra indicate that emission from the outflow is stronger at a radius of ∼10 kpc in high mass galaxies than in low mass galaxies.