Highlights

Evidence for Cored Dark Matter Distributions in Galaxies at z ~ 1–2

Kinematics are a powerful tool to understand galaxies’ structures and mass composition, as they directly trace a galaxy’s entire mass. This is one of the only ways to probe components that do not emit light, such as the dark matter halo. To decompose a galaxy’s total dynamics into the contributions from the bulge, disk, and halo, it is necessary to have deep observations so the kinematics can be traced to large galactic radii.

However, distant galaxies at the epoch of peak cosmic star formation (“cosmic noon”, z ~ 1–3) are faint, so obtaining such deep data requires large amounts of observing time even on the largest, most sensitive telescopes currently in operation. Initial studies focusing on a handful of individual galaxies (Genzel et al. 2017, Übler et al. 2018) and on stacked galaxy profiles (Lang et al. 2017) provided useful first constraints, but analyses of larger samples of individual galaxy kinematics are necessary to examine widespread trends.

We have compiled a sample of 100 star-forming galaxies at high redshift (z ~ 1–2) with deep, spatially resolved observations from multiple sources, including the KMOS^3D, SINS/zC-SINF, and NOEMA^3D surveys as well as additional observations with LBT/LUCI. Using these deep data, we are able to perform such mass modeling to detangle the galaxy and dark matter halo profiles (Genzel et al. 2020, Price et al. 2021).

One of the key findings is that galaxies at cosmic noon (z ~ 1–2) are baryon-dominated on galaxy scales, with higher-mass (or higher-mass surface density) galaxies having lower dark matter fractions. Some of these galaxies have such low f_DM(R_e) that the profiles of their dark matter halos are likely to be cored, in contrast to predictions for more “peaked” halo profiles (i.e., the NFW profile).

Compared to today’s galaxies (z = 0), the RC100 star-forming disk galaxies at z ~ 1 have similar f_DM(R_e) – M_baryon values to the star-forming, disky galaxies, while the RC100 galaxies at z ~ 2 are more like today’s quiescent, elliptical galaxies. These z ~ 2 RC100 galaxies probably represent the progenitor population of today’s elliptical galaxies, with the transformation to quiescence possibly happening not long after the observed epoch, given the overlap with quiescent galaxies seen at z = 1.7 (grey region).

With the ongoing NOEMA^3D survey, work will continue to expand this kinematic modeling to more galaxies. Furthermore, using existing data and upcoming observations with VLT/ERIS, we will begin to explore signatures of mass transport and outflows, to better understand bulge growth and other physical processes regulating galaxy growth.

A rich set of data is available for the RC100 sample, including imaging from the Hubble Space telescope and emission line observations, which provide spatially resolved kinematic maps (upper left panels). For kinematic analysis, we extract 1D profiles along the galaxies’ kinematic major axes (lower left panels). We find that more distant galaxies (z ~ 2; red circles) in the RC100 sample generally have lower dark matter fractions than more nearby galaxies (z ~ 1; blue circles). The z ~ 1 RC100 galaxies are more similar to today’s star-forming, disky "late-type" galaxies (LTGs, including our own Milky Way), while those at z ~ 2 are more similar to today’s quiescent, elliptical “early-type” galaxies (ETGs).

A rich set of data is available for the RC100 sample, including imaging from the Hubble Space telescope and emission line observations, which provide spatially resolved kinematic maps (upper left panels). For kinematic analysis, we extract 1D profiles along the galaxies’ kinematic major axes (lower left panels). We find that more distant galaxies (z ~ 2; red circles) in the RC100 sample generally have lower dark matter fractions than more nearby galaxies (z ~ 1; blue circles). The z ~ 1 RC100 galaxies are more similar to today’s star-forming, disky "late-type" galaxies (LTGs, including our own Milky Way), while those at z ~ 2 are more similar to today’s quiescent, elliptical “early-type” galaxies (ETGs).

2018

Outflow Demographics and Physical Properties at z ~ 1–3

Exploiting our full KMOS^3D and SINS/zC-SINF surveys of near-IR IFU spectroscopy of z ~ 1–3 galaxies, we obtained the most complete census to date of galactic-scale ionized gas outflows at the peak epoch of cosmic star formation and AGN activity. The sample of ~ 600 primarily mass-selected galaxies spans wide ranges in stellar mass and star formation rate; the selection by mass, rather than by properties biased toward star formation (SF) or AGN activity, makes it ideally suited for a population-averaged characterization of winds as relevant to galaxy evolution. Compared to slit spectra, the IFU data greatly facilitate the separation between the broad outflow component in Hα+[NII]+[SII] and the narrower component from SF. Our studies show how outflows driven by SF and AGN are spatially, spectrally, and demographically distinct. SF-driven winds, launched near bright star-forming clumps across disks, have typical speeds of ~ 450 km/s below the hosts’ escape velocity except at log(M_⁎/M_☉) < 10.3; the prevalence of these winds depends on SF properties, not mass. AGN-driven winds originate from the nuclear regions, are ubiquitous in log(M_⁎/M_☉) > 10.7 galaxies hosting a massive bulge but rare at lower masses, irrespective of SF activity; with velocities of ~ 1,500 km/s, they can escape the galaxies. For the first time, the high S/N spectra constrain the density in high-z SF-driven winds from the broad [SII] doublet ratio, yielding n_e~ 400 cm^-3; for AGN-driven winds, a higher n_e~ 1,000 cm^-3 is inferred. These densities are a factor of several higher than previously assumed values, and lead to correspondingly more modest mass outflow rates ~ 0.1–0.4 x SFRs in warm ionized gas. The tension with theoretical work, requiring mass outflow rates ≳ SFRs to reproduce the observed relationships between galaxy mass and metallicity as well as galaxy mass and halo mass at log(M_⁎/M_☉) < 10.7, could be alleviated if substantial mass, momentum, and energy were ejected in hotter and/or colder phases than the ~ 10⁴K ionized gas probed by our data. The fast, high-duty-cycle AGN-driven winds at high masses carry significant energy (~ 1% that of the AGN), which may contribute to heat halo gas and help prevent further gas infall. Our results are consistent with recent EAGLE and Illustris/TNG numerical simulations, which suggest that such a mechanism, acting also at the modest luminosities and Eddington ratios of the majority of the KMOS^3D and SINS/zC-SINF AGN, may be more effective at widespread and long-term quenching than ejective "QSO mode" feedback in rare, high-luminosity, high-Eddington-ratio AGN.

2016

The Angular Momentum Distribution of z ~ 1−3 Star-Forming Galaxies

Top: These panels show an example of a z ~ 1 galaxy observed in KMOS^3D: HST IJH bands color map, KMOS Hα flux, velocity, velocity dispersion maps, and major axis velocity and dispersion profiles. Bottom left: The plot shows the angular momentum parameter distribution from modeling the data assuming exponential baryonic disks in NFW dark matter halos neglecting or accounting for adiabatic contraction (AC), and without AC but accounting for deviations from a pure disk profile for the galaxies. The distributions are log-normal, with a mean value ~0.037 and dispersion in logarithmic units of ~0.2 dex. Bottom right: The plot shows the angular momentum parameter as a function of stellar or gas mass surface density within the half-light radius R_1/2 corrected for the redshift evolution of galaxy sizes, and of stellar mass surface density in the inner 1 kiloparsec.

The angular momentum links galaxies to their host dark matter halos and contains the imprint of their baryonic mass assembly history. We exploited the high-quality, spatially resolved Hα kinematics of a representative subset of 360 log(M_⁎/M_☉) ~ 9.3–11.8 z ~ 1–3 star-forming galaxies from our KMOS^3D and SINS/zC-SINF surveys, obtained with the near-IR multiobject KMOS and AO-assisted single-object SINFONI integral field spectrographs at the Very Large Telescope. That way, we derived for the first time robustly the angular momentum distribution of massive star-forming galaxies around the peak epoch of cosmic star formation. The inferred halo scale angular momentum distribution of the galaxies is consistent with the theoretical prediction for their dark matter halos in terms of mean spin parameter 〈λ〉 ~ 0.037 and dispersion σ(log λ) ~ 0.2. Spin parameters correlate with disk size and stellar surface density but do not depend significantly on halo mass, stellar mass, or redshift. Our data support the long-standing assumption that, on average, the specific angular momentum of disks reflects that of their dark matter halos (j_d= j_DM). The weak correlation between λ×(j_d/j_DM) and stellar surface density in the inner 1 kiloparsec suggests that internal processes lead to "compaction" and dense core formation inside massive high-z disks. The analysis of our sample further yields an average stellar disk-to-dark matter mass ratio of ~ 2%, consistent with abundance matching results. Including the molecular gas, the total baryonic disk-to-dark matter mass ratio is ~ 5% for halos near 10¹² M_☉, which corresponds to 31% of the cosmologically available baryons, implying that high-redshift disks are strongly baryon-dominated.

These results can be found in Burkert et al. 2016.

The Mass Budget of Early Star-Forming Galaxies from KMOS^3D

Left figure, top row: The panels show the distribution of the ratio of stellar to dynamical mass (left) and of baryonic to dynamical mass (right). Here, the baryonic mass estimate accounts for the stellar content as well as molecular gas inferred from state-of-the-art scaling relations based on CO and dust observations (from Genzel et al. 2015). Left figure, bottom row: The panels show the correlation between dynamic, stellar, and baryonic masses, and indicate that when accounting for the gas, baryons strongly dominate the total enclosed mass. Right figure: Stellar mass fractions correlate with stellar mass surface density. Qualitatively similar relations are observed for the baryonic mass fractions, also as a function of gas and total dynamical mass surface densities, and follow from the self-consistent modeling of stars, gas, and dark matter in a ΛCDM context by the Illustris cosmological hydrodynamical simulation (e.g., Vogelsberger et al. 2014; Genel et al. 2015).

Left figure, top row: The panels show the distribution of the ratio of stellar to dynamical mass (left) and of baryonic to dynamical mass (right). Here, the baryonic mass estimate accounts for the stellar content as well as molecular gas inferred from state-of-the-art scaling relations based on CO and dust observations (from Genzel et al. 2015). Left figure, bottom row: The panels show the correlation between dynamic, stellar, and baryonic masses, and indicate that when accounting for the gas, baryons strongly dominate the total enclosed mass.
Right figure: Stellar mass fractions correlate with stellar mass surface density. Qualitatively similar relations are observed for the baryonic mass fractions, also as a function of gas and total dynamical mass surface densities, and follow from the self-consistent modeling of stars, gas, and dark matter in a ΛCDM context by the Illustris cosmological hydrodynamical simulation (e.g., Vogelsberger et al. 2014; Genel et al. 2015).

We exploited our deep integral-field spectroscopic observations from KMOS^3D to dynamically constrain the mass budget of 240 star-forming disks at 0.6 < z < 2.6. Our sample consists of massive (≳ 10^9.8M_⊙) galaxies with sizes R_e ≳ 2 kiloparsec. By contrasting the observed velocity and velocity dispersion profiles to dynamical models, we find that on average the stellar content contributes about 32%, and the total (stellar and gas) baryonic content amounts to about 56% of the dynamical mass budget. Nearly all disks at z > 2 are strongly baryon-dominated within their half-light radius. Substantial object-to-object variations in both stellar and baryonic mass fractions are observed, correlating most strongly with measures of surface density. Our findings can be interpreted as more extended disks probing further (and more compact disks probing less far) into the dark matter halos that host them.

These results can be found in Wuyts et al. 2016, ApJ 831, 149.

Consistent Evolution of Metallicity and Metallicity Gradients from z ~ 2.7 to z ~ 0.6

These panels illustrate the mass-metallicity relationship in three redshift bins derived from the integrated [NII]/Hα line ratio, using the Pettini & Pagel (2004) calibration. The data are shown for stacked spectra of star-forming galaxies binned in stellar mass from KMOS3D (large black and blue symbols) and SINS/zC-SINF+LUCI (large orange symbols). Small gray dots indicate individual measurements from KMOS3D. Excluding galaxies with an AGN (shown with diamonds) reduces the sensitivity at high masses but otherwise results in similar relationships. The dotted line in all panels shows the z ~ 0.8 relationship by Kewley & Ellison (2008). In the middle panel, the dashed line represents the relationship derived from the FMOS survey at z ~ 1.6 by Zahid et al. (2014). In the right panel, the different dashed lines plot those derived at z ~ 2.3 from Erb et al. (2006), Steidel et al. (2014; KBSS survey), and Sanders et al. (2015; MOSDEF survey). — These panels illustrate the mass-metallicity relationship in three redshift bins derived from the integrated [NII]/Hα line ratio, using the Pettini & Pagel (2004) calibration. The data are shown for stacked spectra of star-forming galaxies binned in stellar mass from KMOS^3D (large black and blue symbols) and SINS/zC-SINF+LUCI (large orange symbols). Small gray dots indicate individual measurements from KMOS^3D. Excluding galaxies with an AGN (shown with diamonds) reduces the sensitivity at high masses but otherwise results in similar relationships. The dotted line in all panels shows the z ~ 0.8 relationship by Kewley & Ellison (2008). In the middle panel, the dashed line represents the relationship derived from the FMOS survey at z ~ 1.6 by Zahid et al. (2014). In the right panel, the different dashed lines plot those derived at z ~ 2.3 from Erb et al. (2006), Steidel et al. (2014; KBSS survey), and Sanders et al. (2015; MOSDEF survey).

These panels illustrate the mass-metallicity relationship in three redshift bins derived from the integrated [NII]/Hα line ratio, using the Pettini & Pagel (2004) calibration. The data are shown for stacked spectra of star-forming galaxies binned in stellar mass from KMOS^3D (large black and blue symbols) and SINS/zC-SINF+LUCI (large orange symbols). Small gray dots indicate individual measurements from KMOS^3D. Excluding galaxies with an AGN (shown with diamonds) reduces the sensitivity at high masses but otherwise results in similar relationships. The dotted line in all panels shows the z ~ 0.8 relationship by Kewley & Ellison (2008). In the middle panel, the dashed line represents the relationship derived from the FMOS survey at z ~ 1.6 by Zahid et al. (2014). In the right panel, the different dashed lines plot those derived at z ~ 2.3 from Erb et al. (2006), Steidel et al. (2014; KBSS survey), and Sanders et al. (2015; MOSDEF survey).

This plot shows the iInferred metallicity evolution for M ~ 1010 M☉ from KMOS3D across the wide redshift interval from z ~ 2.3 to z ~ 0.9 probed consistently by our YJ-, H-, and K-band observations (large black diamonds). Other relevant measurements in the literature are plotted as circles after recalibration onto the same metallicity system of Maiolino et al. (Kewley & Ellison 2008, Zahid et al. 2014, Steidel et al. 2014, Sanders et al. 2015, Erb et al. 2006, Onodera et al. 2016, Maiolino et al. 2008). The KMOS3D match the expected evolution following the "equilibrium growth model" of Lilly et al. 2013 remarkably well, assuming the local fundamental relation between M⁎, metallicity, and SFR of Mannucci et al. 2010 and an evolving star formation efficiency ∝ (1+z)0.34±0.15 of Genzel et al. 2015. — This plot shows the iInferred metallicity evolution for M ~ 10¹⁰ M_☉ from KMOS^3D across the wide redshift interval from z ~ 2.3 to z ~ 0.9 probed consistently by our YJ-, H-, and K-band observations (large black diamonds). Other relevant measurements in the literature are plotted as circles after recalibration onto the same metallicity system of Maiolino et al. (Kewley & Ellison 2008, Zahid et al. 2014, Steidel et al. 2014, Sanders et al. 2015, Erb et al. 2006, Onodera et al. 2016, Maiolino et al. 2008). The KMOS^3D match the expected evolution following the "equilibrium growth model" of Lilly et al. 2013 remarkably well, assuming the local fundamental relation between *M_⁎*, metallicity, and SFR of Mannucci et al. 2010 and an evolving star formation efficiency ∝ (1+z)^0.34±0.15 of Genzel et al. 2015.

This plot shows the iInferred metallicity evolution for M ~ 10¹⁰ M_☉ from KMOS^3D across the wide redshift interval from z ~ 2.3 to z ~ 0.9 probed consistently by our YJ-, H-, and K-band observations (large black diamonds). Other relevant measurements in the literature are plotted as circles after recalibration onto the same metallicity system of Maiolino et al. (Kewley & Ellison 2008, Zahid et al. 2014, Steidel et al. 2014, Sanders et al. 2015, Erb et al. 2006, Onodera et al. 2016, Maiolino et al. 2008). The KMOS^3D match the expected evolution following the "equilibrium growth model" of Lilly et al. 2013 remarkably well, assuming the local fundamental relation between *M_⁎*, metallicity, and SFR of Mannucci et al. 2010 and an evolving star formation efficiency ∝ (1+z)^0.34±0.15 of Genzel et al. 2015.

The inferred metallicity gradients from resolved [NII]/Hα radial profiles from KMOS3D (black symbols) are on average flat (as found in previous studies with much smaller and more biased samples) and consistent with theoretical expectations when strong feedback is involved, based on numerical simulations (colored lines; from Pilkington et al. 2012 based on Rahimi et al. 2011, Kobayashi & Nakasato 2012, Few et al. 2012, Gibson et al. 2013). — The inferred metallicity gradients from resolved [NII]/Hα radial profiles from KMOS^3D (black symbols) are on average flat (as found in previous studies with much smaller and more biased samples) and consistent with theoretical expectations when strong feedback is involved, based on numerical simulations (colored lines; from Pilkington et al. 2012 based on Rahimi et al. 2011, Kobayashi & Nakasato 2012, Few et al. 2012, Gibson et al. 2013).

The inferred metallicity gradients from resolved [NII]/Hα radial profiles from KMOS^3D (black symbols) are on average flat (as found in previous studies with much smaller and more biased samples) and consistent with theoretical expectations when strong feedback is involved, based on numerical simulations (colored lines; from Pilkington et al. 2012 based on Rahimi et al. 2011, Kobayashi & Nakasato 2012, Few et al. 2012, Gibson et al. 2013).

We used the [NII]λ6584/Hα ratio as probe of the gas-phase oxygen abundance in over 400 galaxies representative of the bulk of the star-forming population from our KMOS^3D and SINS/zC-SINF surveys with KMOS and SINFONI at the Very Large Telescope, and LUCI sample at the Large Binocular Telescope. We constructed statistically robust mass-metallicity relationships determined consistently from the same indicator over a wide redshift range spanning z = 0.6–2.7. We found no significant dependence of the inferred metallicity on star formation rate (SFR) at fixed redshift and mass; this result, most significant for the z ~ 1 subsample, is in contrast to findings at z ~ 0 that led to the proposed "fundamental metallicity relation," whereby lower metallicities in high-z galaxies would result naturally from their elevated SFRs. With the spatially resolved KMOS^3D and SINS/zC-SINF data, we derived abundance gradients in ~ 200 galaxies, tripling current literature samples. The gradients are on average flat, with only ~ 10% of them having a slope significantly offset from zero even when accounting for beam smearing. Given that most of the galaxies show no sign of interaction/merging, these results suggest efficient metal mixing mediated by strong outflows as predicted by cosmological simulations, and observed in a majority of our sample (driven by vigorous star formation and by AGN). Alternatively, shocks and ionization effects could contribute to mimic flat line ratio gradients.

These results can be found in Wuyts et al. 2016, ApJ, 827, 74.
Earlier results based on the first-year KMOS^3D data, and the SINS-zC-SINF and LBT/LUCI surveys appeared in Wuyts et al. 2014, ApJ, 789, 40.

2015

Successful Start of the KMOS^3D Survey

This image illustrates the Hα velocity fields for 250 galaxies at z ~ 0.9 and z ~ 2.2 from our KMOS3D survey. The velocity fields are shown on the same angular scale, and blue to red colors correspond to blueshifted to redshifted velocities relative to the systemic velocity of each source. The galaxies are plotted within 0.1 dex of their location in the stellar mass − star formation rate (SFR) plane, where the SFRs are normalized to that of the main sequence ("MS") at log(M⁎/M☉) = 10.5, and at the median z = 0.9 and 2.2 for the galaxies in the 0.7 < z < 1.1 and 1.9 < z < 2.7 intervals (observed in the YJ-and K-bands, respectively). A majority of star-forming galaxies are disks, reflected in their smooth monotonically varying velocity gradients. Several resolved disks are even uncovered among the submain sequence population at high masses. — This image illustrates the Hα velocity fields for 250 galaxies at z ~ 0.9 and z ~ 2.2 from our KMOS^3D survey. The velocity fields are shown on the same angular scale, and blue to red colors correspond to blueshifted to redshifted velocities relative to the systemic velocity of each source. The galaxies are plotted within 0.1 dex of their location in the stellar mass − star formation rate (SFR) plane, where the SFRs are normalized to that of the main sequence ("MS") at log(M_⁎/M_☉) = 10.5, and at the median z = 0.9 and 2.2 for the galaxies in the 0.7 < z < 1.1 and 1.9 < z < 2.7 intervals (observed in the YJ-and K-bands, respectively). A majority of star-forming galaxies are disks, reflected in their smooth monotonically varying velocity gradients. Several resolved disks are even uncovered among the submain sequence population at high masses.

This image illustrates the Hα velocity fields for 250 galaxies at z ~ 0.9 and z ~ 2.2 from our KMOS^3D survey. The velocity fields are shown on the same angular scale, and blue to red colors correspond to blueshifted to redshifted velocities relative to the systemic velocity of each source. The galaxies are plotted within 0.1 dex of their location in the stellar mass − star formation rate (SFR) plane, where the SFRs are normalized to that of the main sequence ("MS") at log(M_⁎/M_☉) = 10.5, and at the median z = 0.9 and 2.2 for the galaxies in the 0.7 < z < 1.1 and 1.9 < z < 2.7 intervals (observed in the YJ-and K-bands, respectively). A majority of star-forming galaxies are disks, reflected in their smooth monotonically varying velocity gradients. Several resolved disks are even uncovered among the submain sequence population at high masses.

The compilation of measurements of the disk-intrinsic velocity dispersions σ0 from KMOS3D and other near-IR IFU surveys at z > 0.7, and optical long-slit or imaging spectroscopic surveys at z ~ 0 shows an increase with redshift following σ0 ∝ 1+z. This evolution is consistent with theoretical expectations for marginally stable gas-rich disks given the observed evolution of molecular gas fractions out to z ~ 3 and the galaxies' distribution in rotation velocities. — The compilation of measurements of the disk-intrinsic velocity dispersions σ₀ from KMOS^3D and other near-IR IFU surveys at z > 0.7, and optical long-slit or imaging spectroscopic surveys at z ~ 0 shows an increase with redshift following σ₀ ∝ 1+z. This evolution is consistent with theoretical expectations for marginally stable gas-rich disks given the observed evolution of molecular gas fractions out to z ~ 3 and the galaxies' distribution in rotation velocities.

The compilation of measurements of the disk-intrinsic velocity dispersions σ₀ from KMOS^3D and other near-IR IFU surveys at z > 0.7, and optical long-slit or imaging spectroscopic surveys at z ~ 0 shows an increase with redshift following σ₀ ∝ 1+z. This evolution is consistent with theoretical expectations for marginally stable gas-rich disks given the observed evolution of molecular gas fractions out to z ~ 3 and the galaxies' distribution in rotation velocities.

Taking advantage of the new and efficient near-IR 24-IFU KMOS instrument, built by a consortium involving MPE, we began the KMOS^3D survey in November 2013, an ambitious and highly successful 75-night GTO program led by a team from MPE IR/Submm, MPE OPINAS, and USM. KMOS^3D is mapping the Hα+[NII]+[SII] emission of 600+ mass-selected galaxies at z ~ 0.6–2.7. The survey is carried out in well-studied extragalactic fields with extensive multi-wavelength data, including the far-IR Herschel PEP survey led by our group, and high-resolution optical/near-IR grism and imaging data from the 3D-HST/CANDELS HST Treasury programs. KMOS^3D is designed to provide an unbiased census from deep integrations (~ 5 to 25 hours) of the same spectral diagnostics resolved on seeing-limited scales of 4 to 5 kiloparsec, over a wide range of galaxy parameters and 5 Gyr of cosmic time. The strategy is uniquely enabling faint-line emission mapping in individual objects and pushing IFU studies into new regimes, such as lower-mass main-sequence star-forming galaxies, and high-mass submain-sequence galaxies in the process of quenching. Altogether, KMOS^3D spans two orders of magnitude in stellar mass (log [M_⁎/M_☉] ~ 9.5–11.5) and three orders of magnitude in SFR relative to the main sequence (SFR/SFR_MS ~ 0.01–10). KMOS^3D now confirmed robustly our earlier results from SINS/zC-SINF on the kinematics and structure of high-z star-forming galaxies. The dynamical support of at least 70% of all z ~ 1–3 massive star-forming galaxies is dominated by ordered disk rotation, unlike what would be expected in the case of frequent (major) merging. However, the high-z disks differ significantly from nearby spiral galaxies: the measured large local random motions from Hα emission (σ₀ ≳ 25 km/s) reveal turbulent ionized gas disks. The disk velocity dispersion increases with redshift as σ₀∝ (1+z), in line with expectations for gas-rich disks and the observed evolution in cold gas mass fractions. By targeting the same spectral diagnostics of homogeneously selected samples, observed and analyzed in the same way, KMOS^3D is providing the most consistent IFU study of the evolution of resolved kinematics, star formation, and warm ISM properties of z ~ 0.7–2.7 galaxies.

The KMOS^3D survey design and strategy, and first-year results appeared in Wisnioski et al. 2015, ApJ, 799, 209.

2014

Widespread AGN-Driven Outflows in the Most Massive z ~ 1−2.5 Star-Forming Galaxies

Top left: The stacked spectrum of the central regions of log(M⁎/M☉) > 10.9 galaxies at z ~ 0.7−2.7 from our KMOS3D and SINS/zC-SINF surveys shows, at high S/N, the broad (FWHM ~ 500−2,000 km/s) emission component in Hα, [NII], and [SII] associated with AGN-driven outflows. It contrasts with the stacked spectrum of the outer disk regions of the same galaxies, which exhibits a weaker and narrower (FWHM ~ 400 km/s) broad emission component from star-formation-driven outflows. Bottom left: For five of the six log(M⁎/M☉) > 10.9 SINS/zC-SINF galaxies with deep AO-assisted SINFONI observations at ~ 0.2 arcseconds resolution, the broad emission originates primarily from the center, where a stellar bulge is seen in the rest-optical light and derived stellar mass distribution. The nuclear broad emission is resolved in most cases and implies an extent of 2 to 3 kiloparsec. For BX610, the broadest outflow emission is seen around the center; strong but narrower outflow emission is also detected at the location of the bright star-forming "clump" to the southwest. For each galaxy, a pair of images is shown: the left one displays the rest-optical stellar continuum light and star formation (observed H band map from HST or extracted from the SINFONI data in red colors, HST J-band map in blue colors when available, and narrow star-formation-dominated Hα line emission from SINFONI in green colors), and the right one shows the same H-band and narrow Hα maps (in red and green, respectively) with the broad, outflow-dominated Hα+[NII] emission component overlaid in white contours. Top right: This plot shows variations in broad component velocity FWHM with galaxy stellar mass for the stacked spectra of the inner 2 to 3 kiloparsec (large circles and red-shaded area) and outer disk regions (squares and green-shaded area) of our z ~ 0.7−2.7 star-forming galaxies. Spectra were stacked per mass bin and separately for galaxies with specific star formation rate above (red circles and dark-green squares) and below (blue circles and light green squares) the main sequence. Measurements from individual nuclear spectra with best S/N are overplotted at star symbols; the three galaxies dominated by BLR emission are labeled in the plot, and excluded from the stacks. The onset of the broad nuclear emission associated with AGN-driven outflows is very sharp toward high masses, while star-formation-driven outflows with more modest velocities are detected across disks at all masses. Bottom right: The incidence of broad nuclear emission (red circles) increases sharply above log(M*/M☉) ~ 10.9, reaching ~2/3 at the highest masses. Although the fraction of AGN among our samples (yellow stars) or in the general galaxy population at similar redshifts in the GOODS and COSMOS extragalactic survey fields (gray- and green-shaded polygons) follows a similar trend, the frequency is < 50%. The nuclear outflow phenomenon thus has a higher duty cycle than the highly variable AGN emission. — Top left: The stacked spectrum of the central regions of log(M_⁎/M_☉) > 10.9 galaxies at z ~ 0.7−2.7 from our KMOS^3D and SINS/zC-SINF surveys shows, at high S/N, the broad (FWHM ~ 500−2,000 km/s) emission component in Hα, [NII], and [SII] associated with AGN-driven outflows. It contrasts with the stacked spectrum of the outer disk regions of the same galaxies, which exhibits a weaker and narrower (FWHM ~ 400 km/s) broad emission component from star-formation-driven outflows. Bottom left: For five of the six log(M_⁎/M_☉) > 10.9 SINS/zC-SINF galaxies with deep AO-assisted SINFONI observations at ~ 0.2 arcseconds resolution, the broad emission originates primarily from the center, where a stellar bulge is seen in the rest-optical light and derived stellar mass distribution. The nuclear broad emission is resolved in most cases and implies an extent of 2 to 3 kiloparsec. For BX610, the broadest outflow emission is seen around the center; strong but narrower outflow emission is also detected at the location of the bright star-forming "clump" to the southwest. For each galaxy, a pair of images is shown: the left one displays the rest-optical stellar continuum light and star formation (observed H band map from HST or extracted from the SINFONI data in red colors, HST J-band map in blue colors when available, and narrow star-formation-dominated Hα line emission from SINFONI in green colors), and the right one shows the same H-band and narrow Hα maps (in red and green, respectively) with the broad, outflow-dominated Hα+[NII] emission component overlaid in white contours. Top right: This plot shows variations in broad component velocity FWHM with galaxy stellar mass for the stacked spectra of the inner 2 to 3 kiloparsec (large circles and red-shaded area) and outer disk regions (squares and green-shaded area) of our z ~ 0.7−2.7 star-forming galaxies. Spectra were stacked per mass bin and separately for galaxies with specific star formation rate above (red circles and dark-green squares) and below (blue circles and light green squares) the main sequence. Measurements from individual nuclear spectra with best S/N are overplotted at star symbols; the three galaxies dominated by BLR emission are labeled in the plot, and excluded from the stacks. The onset of the broad nuclear emission associated with AGN-driven outflows is very sharp toward high masses, while star-formation-driven outflows with more modest velocities are detected across disks at all masses. Bottom right: The incidence of broad nuclear emission (red circles) increases sharply above log(M_*/M_☉) ~ 10.9, reaching ~2/3 at the highest masses. Although the fraction of AGN among our samples (yellow stars) or in the general galaxy population at similar redshifts in the GOODS and COSMOS extragalactic survey fields (gray- and green-shaded polygons) follows a similar trend, the frequency is < 50%. The nuclear outflow phenomenon thus has a higher duty cycle than the highly variable AGN emission.

Top left: The stacked spectrum of the central regions of log(M_⁎/M_☉) > 10.9 galaxies at z ~ 0.7−2.7 from our KMOS^3D and SINS/zC-SINF surveys shows, at high S/N, the broad (FWHM ~ 500−2,000 km/s) emission component in Hα, [NII], and [SII] associated with AGN-driven outflows. It contrasts with the stacked spectrum of the outer disk regions of the same galaxies, which exhibits a weaker and narrower (FWHM ~ 400 km/s) broad emission component from star-formation-driven outflows. Bottom left: For five of the six log(M_⁎/M_☉) > 10.9 SINS/zC-SINF galaxies with deep AO-assisted SINFONI observations at ~ 0.2 arcseconds resolution, the broad emission originates primarily from the center, where a stellar bulge is seen in the rest-optical light and derived stellar mass distribution. The nuclear broad emission is resolved in most cases and implies an extent of 2 to 3 kiloparsec. For BX610, the broadest outflow emission is seen around the center; strong but narrower outflow emission is also detected at the location of the bright star-forming "clump" to the southwest. For each galaxy, a pair of images is shown: the left one displays the rest-optical stellar continuum light and star formation (observed H band map from HST or extracted from the SINFONI data in red colors, HST J-band map in blue colors when available, and narrow star-formation-dominated Hα line emission from SINFONI in green colors), and the right one shows the same H-band and narrow Hα maps (in red and green, respectively) with the broad, outflow-dominated Hα+[NII] emission component overlaid in white contours. Top right: This plot shows variations in broad component velocity FWHM with galaxy stellar mass for the stacked spectra of the inner 2 to 3 kiloparsec (large circles and red-shaded area) and outer disk regions (squares and green-shaded area) of our z ~ 0.7−2.7 star-forming galaxies. Spectra were stacked per mass bin and separately for galaxies with specific star formation rate above (red circles and dark-green squares) and below (blue circles and light green squares) the main sequence. Measurements from individual nuclear spectra with best S/N are overplotted at star symbols; the three galaxies dominated by BLR emission are labeled in the plot, and excluded from the stacks. The onset of the broad nuclear emission associated with AGN-driven outflows is very sharp toward high masses, while star-formation-driven outflows with more modest velocities are detected across disks at all masses. Bottom right: The incidence of broad nuclear emission (red circles) increases sharply above log(M_*/M_☉) ~ 10.9, reaching ~2/3 at the highest masses. Although the fraction of AGN among our samples (yellow stars) or in the general galaxy population at similar redshifts in the GOODS and COSMOS extragalactic survey fields (gray- and green-shaded polygons) follows a similar trend, the frequency is < 50%. The nuclear outflow phenomenon thus has a higher duty cycle than the highly variable AGN emission.

Following the detection of powerful star-formation-driven ionized gas outflows originating throughout the disk regions and especially around intensely star-forming clumps in our z ~ 2 SINS/zC-SINF sample, new observations with SINFONI+AO and KMOS uncovered distinct high-velocity outflows in the centers of the most massive but otherwise normal star-forming galaxies. With a FWHM ~ 500−2,000 km/s seen in Hα as well as in forbidden [NII]λλ6548,6584 and [SII]λλ6716,6731 line emission, elevated [NII]/Hα ratio > 0.5, and an extent of 2−3 kiloparsec derived from five sources with high-resolution AO-assisted observations, this broad emission component is most plausibly originating from AGN-driven outflows. The frequency of these nuclear outflows rises sharply at log(M_⁎/M_☉) ~ 10.9, reaching 2/3 above this mass. These star-forming galaxies were selected based on mass and on their location around the main sequence of star-forming galaxies, rather than by the presence of an AGN. In fact, < 50% of these galaxies are classified as hosting an AGN from classical X-ray, optical, IR, and radio indicators, suggesting the nuclear outflows have a higher duty cycle than the extremely variable AGN activity. The typical high inferred mass outflow rates (dM_out/dt > SFR) and momentum deposition rates (v_out×dM_out/dt ~ 20×L/c), together with the presence of massive bulges, and with evidence for suppressed star formation and gravitational quenching in the inner 2 to 3 kiloparsec of half of the galaxies, make a compelling case that these nuclear winds play an important role in clearing the central regions of gas prior to quenching.

These results appeared in Förster Schreiber et al. 2014, ApJ, 787, 37 and Genzel et al. 2014, ApJ, 796, 7.

Bulge Growth and Quenching since z = 2.5

This plot shows the bulge-to-total (B/T) stellar mass ratio as a function of the total stellar mass of ~ 6,800 galaxies at 1010M⊙ and 0.5 < z < 2.5. The B/T of each galaxy was derived by fitting a disk+bulge model to the stellar mass maps obtained by modeling the spatially resolved rest-UV+optical spectral energy distributions from the multiband CANDELS HST imaging. The whole sample (black nonfilled symbols) is divided into star-forming (blue symbols) and quiescent galaxies (red symbols). The respective shaded areas mark the 50th percentile scatter of the distributions within one bin, while the error bars indicate the uncertainty on the median value. Results for the lower and upper halves of the redshift interval are marked with dotted and dashed lines, respectively. Star-forming galaxies show a clear trend of increasing B/T with increasing stellar mass, reaching B/T ~ 0.4–0.5 above 1011 M☉ and reflecting the buildup of central stellar mass concentrations in main-sequence galaxies up to z ~ 2.5. — This plot shows the bulge-to-total (B/T) stellar mass ratio as a function of the total stellar mass of ~ 6,800 galaxies at 10¹⁰M_⊙ and 0.5 < z < 2.5. The B/T of each galaxy was derived by fitting a disk+bulge model to the stellar mass maps obtained by modeling the spatially resolved rest-UV+optical spectral energy distributions from the multiband CANDELS HST imaging. The whole sample (black nonfilled symbols) is divided into star-forming (blue symbols) and quiescent galaxies (red symbols). The respective shaded areas mark the 50th percentile scatter of the distributions within one bin, while the error bars indicate the uncertainty on the median value. Results for the lower and upper halves of the redshift interval are marked with dotted and dashed lines, respectively. Star-forming galaxies show a clear trend of increasing B/T with increasing stellar mass, reaching B/T ~ 0.4–0.5 above 10¹¹ M_☉ and reflecting the buildup of central stellar mass concentrations in main-sequence galaxies up to z ~ 2.5.

This plot shows the bulge-to-total (B/T) stellar mass ratio as a function of the total stellar mass of ~ 6,800 galaxies at 10¹⁰M_⊙ and 0.5 < z < 2.5. The B/T of each galaxy was derived by fitting a disk+bulge model to the stellar mass maps obtained by modeling the spatially resolved rest-UV+optical spectral energy distributions from the multiband CANDELS HST imaging. The whole sample (black nonfilled symbols) is divided into star-forming (blue symbols) and quiescent galaxies (red symbols). The respective shaded areas mark the 50th percentile scatter of the distributions within one bin, while the error bars indicate the uncertainty on the median value. Results for the lower and upper halves of the redshift interval are marked with dotted and dashed lines, respectively. Star-forming galaxies show a clear trend of increasing B/T with increasing stellar mass, reaching B/T ~ 0.4–0.5 above 10¹¹ M_☉ and reflecting the buildup of central stellar mass concentrations in main-sequence galaxies up to z ~ 2.5.

Exploiting the deep high-resolution imaging of all five fields part of the HST CANDELS imaging survey, and accurate redshift information provided by the 3D-HST grism survey in the same areas, we investigated the relation between structure and stellar populations for a mass-selected sample of 6,764 galaxies above 10¹⁰M_⊙, spanning the redshift range 0.5 < z < 2.5. For the first time, we fitted two-dimensional models comprising a single Sérsic fit and two-component (i.e., bulge+disk) decompositions not only to the H-band light distributions, but also to the stellar mass maps reconstructed from resolved stellar population modeling. The results confirm that the increased bulge prominence among quiescent galaxies, as reported previously based on rest-optical observations, remains in place when considering the distributions of stellar mass. Moreover, we observed an increase of the typical Sérsic index and bulge-to-total ratio (with median B/T reaching 40 to 50%) among star-forming galaxies above 10¹¹M_⊙. Given that quenching for these most massive systems is likely to be imminent, our findings suggest that significant bulge growth precedes a departure from the star-forming main sequence. We demonstrated that the bulge mass (and ideally knowledge of the bulge and total mass) is a more reliable predictor of the star-forming vs. quiescent state of a galaxy than the total stellar mass. The same trends are predicted by the state-of-the-art, semi-analytic model by Somerville et al. Here, bulges and black holes grow hand in hand through merging and/or disk instabilities, and feedback from active galactic nuclei shuts off star formation. Further observations will be required to pin down star-formation-quenching mechanisms, but our results imply that they must be internal to the galaxies and closely associated with bulge growth.

These results appeared in Lang et al. 2014, ApJ, 788, 11.

Evidence for Gravitational Quenching from SINS/zC-SINF

Left: This panel shows the integrated Hα line maps of 19 well-resolved star-forming galaxies from our SINS/zC-SINF AO sample with deep on-source integrations. The galaxies are plotted at their location in the stellar mass vs. star formation rate (SFR) plane, on the same angular scale and with colors scaling linearly with surface brightness; the FWHM angular resolution of these maps is ~ 0.24 arcseconds. The white solid line shows the location of the "main sequence" of z ~ 2 star-forming galaxies assuming a slope of unity, and the dashed lines indicate its 2σ scatter (±0.6 dex in log[SFR]). Many galaxies exhibit ring-like distributions in Hα, especially frequent toward more massive galaxies. In combination with the SINFONI kinematic maps and existing HST J- and H-band maps at similar resolution, the data are indicative of the presence of increasingly massive bulges and suppressed star formation in the central few kiloparsecs of the more massive galaxies. Right: The plot illustrates derived radial profiles of the Toomre Q parameter of the galaxies, color-coded as a function of dynamical mass as follows: log(Mdyn/M☉) = 10.36−10.50 in blue, 10.68−10.93 in green, 11.04−11.28 in orange, and 11.34−11.41 in red. The gray-shaded interval corresponds to the typical resolution element for the sample. All profiles exhibit an increase in Q toward their central regions, with values significantly in excess of the threshold around unity, the more so for the most massive galaxies, suggesting that their bulges may stabilize the gas against gravitational collapse in the inner few kiloparsecs. — Left: This panel shows the integrated Hα line maps of 19 well-resolved star-forming galaxies from our SINS/zC-SINF AO sample with deep on-source integrations. The galaxies are plotted at their location in the stellar mass vs. star formation rate (SFR) plane, on the same angular scale and with colors scaling linearly with surface brightness; the FWHM angular resolution of these maps is ~ 0.24 arcseconds. The white solid line shows the location of the "main sequence" of z ~ 2 star-forming galaxies assuming a slope of unity, and the dashed lines indicate its 2σ scatter (±0.6 dex in log[SFR]). Many galaxies exhibit ring-like distributions in Hα, especially frequent toward more massive galaxies. In combination with the SINFONI kinematic maps and existing HST J- and H-band maps at similar resolution, the data are indicative of the presence of increasingly massive bulges and suppressed star formation in the central few kiloparsecs of the more massive galaxies. Right: The plot illustrates derived radial profiles of the Toomre Q parameter of the galaxies, color-coded as a function of dynamical mass as follows: log(M_dyn/M_☉) = 10.36−10.50 in blue, 10.68−10.93 in green, 11.04−11.28 in orange, and 11.34−11.41 in red. The gray-shaded interval corresponds to the typical resolution element for the sample. All profiles exhibit an increase in Q toward their central regions, with values significantly in excess of the threshold around unity, the more so for the most massive galaxies, suggesting that their bulges may stabilize the gas against gravitational collapse in the inner few kiloparsecs.

Left: This panel shows the integrated Hα line maps of 19 well-resolved star-forming galaxies from our SINS/zC-SINF AO sample with deep on-source integrations. The galaxies are plotted at their location in the stellar mass vs. star formation rate (SFR) plane, on the same angular scale and with colors scaling linearly with surface brightness; the FWHM angular resolution of these maps is ~ 0.24 arcseconds. The white solid line shows the location of the "main sequence" of z ~ 2 star-forming galaxies assuming a slope of unity, and the dashed lines indicate its 2σ scatter (±0.6 dex in log[SFR]). Many galaxies exhibit ring-like distributions in Hα, especially frequent toward more massive galaxies. In combination with the SINFONI kinematic maps and existing HST J- and H-band maps at similar resolution, the data are indicative of the presence of increasingly massive bulges and suppressed star formation in the central few kiloparsecs of the more massive galaxies.
Right: The plot illustrates derived radial profiles of the Toomre Q parameter of the galaxies, color-coded as a function of dynamical mass as follows: log(M_dyn/M_☉) = 10.36−10.50 in blue, 10.68−10.93 in green, 11.04−11.28 in orange, and 11.34−11.41 in red. The gray-shaded interval corresponds to the typical resolution element for the sample. All profiles exhibit an increase in Q toward their central regions, with values significantly in excess of the threshold around unity, the more so for the most massive galaxies, suggesting that their bulges may stabilize the gas against gravitational collapse in the inner few kiloparsecs.

We analyzed the radial distributions of Hα surface brightness, stellar mass surface density, and dynamical mass at ~ 2 kiloparsecs resolution in 19 z ~ 2 star-forming disks with deep AO-assisted SINFONI imaging spectroscopy from our SINS/zC-SINF survey. From the combination of the kinematic maps and the molecular gas mass surface densities inferred from the star formation rate distributions, we derived the radial profiles in Toomre Q parameter for these main-sequence star-forming galaxies, which span about two orders of magnitude in stellar mass (log[M_⁎/M_☉] = 9.6−11.5). In more than half of these galaxies, the Hα distributions cannot be fit by a centrally peaked distribution, such as an exponential, but are better described by a ring or the combination of a ring and an exponential. At the same time, the kinematics data indicate the presence of a mass distribution more centrally concentrated than a single exponential disk component for 5 of the 19 galaxies. The resulting Q profiles are centrally peaked for all, and significantly exceed unity there for ~ 3/4 of the galaxies. The occurrence of Hα rings and of large nuclear Q values appears to be more common for the more massive star-forming galaxies. While the sample is small and biased toward larger sizes, and there remain uncertainties and caveats, the observations are consistent with the "gravitational quenching" scenario, in which cloud fragmentation and global star formation are secularly suppressed in gas-rich high-z disks from the inside out, as the central stellar mass density of the disks grows.

These results appeared in Genzel et al. 2014, ApJ, 785, 75.

2013

The Nature of Dispersion-Dominated Galaxies at High Redshift

Left: This panel shows Hα line maps of the 34 star-forming galaxies from our SINS/zC-SINF AO sample. The top two rows contain the dispersion-dominated objects (with ratio of intrinsic rotation velocity to velocity dispersion vrot/σ0 below ~ 1) and the rest are rotation-dominated. The maps are all plotted on the same angular scale. The typical FWHM resolution of the maps presented here is 0.2 to 0.3 arcseconds (indicated with the red circle). The dispersion-dominated galaxies tend to be more compact than the rotation-dominated ones. Middle and right: These plots illustrate the dependence of vrot and σ0 on the Hα half-light radius R1/2. The measurements for the SINS/zC-SINF galaxies (blue circles) are combined with those of other z ~ 1–2.5 samples observed mostly using AO (Law et al. 2009, 2012 and Wright et al. 2009, red squares; Wisnioski et al. 2011, green circles; Swinbank et al. 2012, cyan upside-down triangles; Épinat et al. 2009, 2012 and Lemoine-Busserolle and Lamareille 2010, black triangles and crosses). The large dark grey symbols show the median values per size bins. The strong trend of velocity vs. size can be well-fit by a linear relation: log(vrot) = 0.62xlog(R1/2)+1.9. In contrast, the velocity dispersion does not appear to show a significant trend. Thus, the trend for smaller galaxies to be dispersion-dominated is in part due to the combination of a possible floor of velocity dispersion, and a linear increase of rotation velocity with size. — Left: This panel shows Hα line maps of the 34 star-forming galaxies from our SINS/zC-SINF AO sample. The top two rows contain the dispersion-dominated objects (with ratio of intrinsic rotation velocity to velocity dispersion v_rot/σ₀ below ~ 1) and the rest are rotation-dominated. The maps are all plotted on the same angular scale. The typical FWHM resolution of the maps presented here is 0.2 to 0.3 arcseconds (indicated with the red circle). The dispersion-dominated galaxies tend to be more compact than the rotation-dominated ones. Middle and right: These plots illustrate the dependence of v_rot and σ₀ on the Hα half-light radius R_1/2. The measurements for the SINS/zC-SINF galaxies (blue circles) are combined with those of other z ~ 1–2.5 samples observed mostly using AO (Law et al. 2009, 2012 and Wright et al. 2009, red squares; Wisnioski et al. 2011, green circles; Swinbank et al. 2012, cyan upside-down triangles; Épinat et al. 2009, 2012 and Lemoine-Busserolle and Lamareille 2010, black triangles and crosses). The large dark grey symbols show the median values per size bins. The strong trend of velocity vs. size can be well-fit by a linear relation: log(v_rot) = 0.62xlog(R_1/2)+1.9. In contrast, the velocity dispersion does not appear to show a significant trend. Thus, the trend for smaller galaxies to be dispersion-dominated is in part due to the combination of a possible floor of velocity dispersion, and a linear increase of rotation velocity with size.

Left: This panel shows Hα line maps of the 34 star-forming galaxies from our SINS/zC-SINF AO sample. The top two rows contain the dispersion-dominated objects (with ratio of intrinsic rotation velocity to velocity dispersion v_rot/σ₀ below ~ 1) and the rest are rotation-dominated. The maps are all plotted on the same angular scale. The typical FWHM resolution of the maps presented here is 0.2 to 0.3 arcseconds (indicated with the red circle). The dispersion-dominated galaxies tend to be more compact than the rotation-dominated ones. Middle and right: These plots illustrate the dependence of v_rot and σ₀ on the Hα half-light radius R_1/2. The measurements for the SINS/zC-SINF galaxies (blue circles) are combined with those of other z ~ 1–2.5 samples observed mostly using AO (Law et al. 2009, 2012 and Wright et al. 2009, red squares; Wisnioski et al. 2011, green circles; Swinbank et al. 2012, cyan upside-down triangles; Épinat et al. 2009, 2012 and Lemoine-Busserolle and Lamareille 2010, black triangles and crosses). The large dark grey symbols show the median values per size bins. The strong trend of velocity vs. size can be well-fit by a linear relation: log(v_rot) = 0.62xlog(R_1/2)+1.9. In contrast, the velocity dispersion does not appear to show a significant trend. Thus, the trend for smaller galaxies to be dispersion-dominated is in part due to the combination of a possible floor of velocity dispersion, and a linear increase of rotation velocity with size.

We analyzed the spatial distributions and kinematics of Hα, [NII], and [SII] emission in 38 star-forming galaxies from our SINS/zC-SINF survey, 34 of which were observed with SINFONI at high resolution using AO. This was supplemented by kinematic data of 43 z ~ 1–2.5 galaxies from the literature. None of these 81 galaxies is an obvious major merger. We found that the kinematic classification of high-redshift galaxies as "dispersion-dominated" or "rotation-dominated" correlates most strongly with their intrinsic sizes. Smaller galaxies are more likely "dispersion-dominated" for two main reasons: 1) The rotation velocity scales linearly with galaxy size, but intrinsic velocity dispersion does not depend on size or may even increase in smaller galaxies, and as such, their ratio is systematically lower for smaller galaxies, and 2) Beam smearing strongly decreases large-scale velocity gradients and increases observed dispersion much more for galaxies with sizes at or below the resolution. Dispersion-dominated galaxies may thus have intrinsic properties similar to the rotation-dominated ones, but are primarily more compact, have lower mass, are less metal-enriched, and may have higher gas fractions, plausibly because they represent an earlier evolutionary state. A key implication of our results is that the derived fraction of dispersion-dominated objects among massive star-forming galaxies at z ~ 1–2.5 is < 20% lower than had been inferred based largely on seeing-limited observations.

These results appeared in Newman et al. 2013, ApJ, 767, 10.

Resolved Star Formation Patterns at 0.7 < z < 1.5

This gallery presents case examples from our sample of 473 massive star-forming galaxies at 0.7 < z < 1.5 in the HST CANDELS/3D-HST extragalactic survey fields. Below the three-color composites, we show the surface brightness distribution in the ACS I (0.8 μm) and WFC3 H (1.6 μm) band, as well as Hα line emission maps extracted from HST grism spectroscopy. The physical resolution in all maps is ~ 1 kiloparsec. Blue, star-forming regions present in the I-band generally dominate the Hα emission as well. Central peaks in surface brightness (i.e., "bulges") appear more prominently in the H-band.

This gallery presents case examples from our sample of 473 massive star-forming galaxies at 0.7 < z < 1.5 in the HST CANDELS/3D-HST extragalactic survey fields. Below the three-color composites, we show the surface brightness distribution in the ACS I (0.8 μm) and WFC3 H (1.6 μm) band, as well as Hα line emission maps extracted from HST grism spectroscopy. The physical resolution in all maps is ~ 1 kiloparsec. Blue, star-forming regions present in the I-band generally dominate the Hα emission as well. Central peaks in surface brightness (i.e., "bulges") appear more prominently in the H-band.

We analyzed the resolved stellar populations in a sample of 473 massive star-forming galaxies at 0.7 < z < 1.5, with multiwavelength broadband imaging from CANDELS and Hα surface brightness profiles at the same kiloparsec resolution from 3D-HST, two HST Treasury extragalactic surveys. Together, this unique data set sheds light on how the assembled stellar mass is distributed within galaxies, and where new stars are being formed. We found the Hα morphologies to resemble more closely those observed in the ACS I (0.8 μm) band than in the WFC3 H (1.6 μm) band, especially for the larger systems. In order to translate the Hα surface brightness profiles to maps of the star formation rate, we derived a novel prescription for Hα dust corrections, which accounts for extra extinction toward HII regions. We found the surface density of star formation to correlate with the surface density of assembled stellar mass within galaxies, akin to the so-called "main sequence" of star formation established on a galaxy-integrated level. Deviations from this relation toward lower equivalent widths are found in the inner regions of galaxies. Clumps and spiral features, on the other hand, are associated with enhanced Hα equivalent widths, bluer colors, and higher specific star formation rates than the underlying disk. Their Hα/UV luminosity ratio is lower than that of the underlying disk, suggesting that the ACS clump selection preferentially picks up those regions of elevated star formation activity that are the least obscured by dust. Our analysis emphasizes that monochromatic studies of galaxy structure can be severely limited by mass-to-light ratio variations due to dust and spatially inhomogeneous star formation histories.

These results appeared in Wuyts et al. 2013, ApJ, 779, 135.

2012

Smoother Stellar Mass Maps and the Longevity of Star-Forming Clumps in High-Redshift Galaxies

Left:This panel shows the fractional contribution of stars younger than 10 Myr and 100 Myr to the total emission at rest-frame far-UV to J-band wavelengths, and to the total mass present in stars. Gray shades indicate Bruzual & Charlot (2003) models with constant star formation histories that started 0.5 Gyr, 1 Gyr, and 2 Gyr prior to the epoch of observation. While the rest-frame V-band provides a better proxy for stellar mass than the UV part of the spectral energy distribution, it is still substantially biased toward the youngest generation of stars. Resolved color information is the key to constraining spatial mass-to-light ratio variations due to nonuniform star formation histories (or dust obscuration) across a galaxy, and recovering the true distribution of stellar mass. Right: These histograms compare structural parameters of concentration and galaxy (ir-)regularity as measured on light maps of different rest-frame wavelength as well as on stellar mass maps reconstructed using pixel-by-pixel stellar population modeling. In mass, galaxies are more compact and smoother than they appear in light, particularly at short wavelengths. This trend is observed over the full redshift range (0.5 < z < 2.5) considered in our analysis.

Left:This panel shows the fractional contribution of stars younger than 10 Myr and 100 Myr to the total emission at rest-frame far-UV to J-band wavelengths, and to the total mass present in stars. Gray shades indicate Bruzual & Charlot (2003) models with constant star formation histories that started 0.5 Gyr, 1 Gyr, and 2 Gyr prior to the epoch of observation. While the rest-frame V-band provides a better proxy for stellar mass than the UV part of the spectral energy distribution, it is still substantially biased toward the youngest generation of stars. Resolved color information is the key to constraining spatial mass-to-light ratio variations due to nonuniform star formation histories (or dust obscuration) across a galaxy, and recovering the true distribution of stellar mass. Right: These histograms compare structural parameters of concentration and galaxy (ir-)regularity as measured on light maps of different rest-frame wavelength as well as on stellar mass maps reconstructed using pixel-by-pixel stellar population modeling. In mass, galaxies are more compact and smoother than they appear in light, particularly at short wavelengths. This trend is observed over the full redshift range (0.5 < z < 2.5) considered in our analysis.

We performed a detailed analysis of the spatially resolved colors and stellar populations of a mass-complete (log(M_⁎/M_☉) > 10) sample of 323 star-forming galaxies at 0.5 < z < 1.5, and 326 star-forming galaxies at 1.5 < z < 2.5 in the ERS and CANDELS-Deep region of the GOODS-South extragalactic field, with very deep imaging from HST. We modeled the seven-band optical ACS and near-IR WFC3 spectral energy distributions of individual bins of pixels, accounting simultaneously for the galaxy-integrated photometric constraints available over a longer wavelength range. We found evidence for redder colors, older stellar ages, and increased dust extinction in the nuclei of galaxies. Large star-forming clumps seen in star formation tracers are less prominent or even invisible on the inferred stellar mass distributions. Our results are consistent with an inside-out disk growth scenario with brief (100 to 200 Myr) episodic local enhancements in star formation superposed on the underlying disk. Alternatively, the young ages of off-center clumps may signal inward clump migration, provided this happens efficiently on the order of an orbital timescale.

These results appeared in Wuyts et al. 2012, ApJ, 753, 114.

Short-Lived Star-Forming Clumps in Cosmological Simulations of z ~ 2 disks: the Impact of Strong Feedback from Massive Stars

The images illustrate the impact of strong feedback from young massive stars in high-resolution cosmological numerical SPH simulations. The time sequence of gas surface density maps shows the disruption of a clump in our model (top), where t = 0 (not shown) is the formation time of the clump. To demonstrate the role of the vigorous wind, it is turned off at z = 2.03 (t = 22 Myr) and the alternative evolution of nondisruption, virialization, and migration is shown for comparison (bottom). The upper rightmost panel shows the mass of gas (solid lines) and young (< 50 Myr) stars (dashed lines) for four clumps as a function of time since their formation. The magenta lines are for the clump highlighted on the top and the black for the clump highlighted on the bottom. The jump in mass of the green lines at t ~ 60 Myr is a result of a merger between two clumps. The typical clump lifetime in the presence of winds is ~ 50 Myr, and the mass of new-formed stars is approximately 10% of the maximum clump gas mass. The mass of new-formed stars internal to the clump decreases following the decrease of the gas mass, as these stars are dispersed out of the clump when the gravitation collapse of the gas is halted by the return to stable conditions with a Toomre Q parameter > 1.

The images illustrate the impact of strong feedback from young massive stars in high-resolution cosmological numerical SPH simulations. The time sequence of gas surface density maps shows the disruption of a clump in our model (top), where t = 0 (not shown) is the formation time of the clump. To demonstrate the role of the vigorous wind, it is turned off at z = 2.03 (t = 22 Myr) and the alternative evolution of nondisruption, virialization, and migration is shown for comparison (bottom). The upper rightmost panel shows the mass of gas (solid lines) and young (< 50 Myr) stars (dashed lines) for four clumps as a function of time since their formation. The magenta lines are for the clump highlighted on the top and the black for the clump highlighted on the bottom. The jump in mass of the green lines at t ~ 60 Myr is a result of a merger between two clumps. The typical clump lifetime in the presence of winds is ~ 50 Myr, and the mass of new-formed stars is approximately 10% of the maximum clump gas mass. The mass of new-formed stars internal to the clump decreases following the decrease of the gas mass, as these stars are dispersed out of the clump when the gravitation collapse of the gas is halted by the return to stable conditions with a Toomre Q parameter > 1.

Many observed massive star-forming z ~ 2 galaxies, including the ones from our SINS/zC-SINF survey, are large disks that exhibit irregular morphologies, with luminous, kiloparsec-sized star-forming clumps. In the framework of turbulent, gas-rich, marginally unstable disks, such clumps form through fragmentation and eventually migrate toward the center of the galaxies where they coalesce to form young bulges. However, our recent findings that clumps are also launching sites of powerful gas outflows that could disrupt them rapidly (highlighted here) raise important questions about their role in the evolution of early disks. To investigate this issue, we used the largest sample to date of high-resolution cosmological smoothed particle hydrodynamics simulations that zoom-in on the formation of individual log(M_⁎/M_☉) ~ 10.5 galaxies in log(M_⁎/M_☉) ~ 12 dark matter halos at z ~ 2. Our code includes strong stellar feedback parameterized as momentum-driven galactic winds. This model reproduces many characteristic features of this observed class of galaxies, such as their clumpy morphologies, smooth and monotonic velocity gradients, high gas fractions (~ 50%), and high specific star formation rates (~ 1 Gyr^-1). In accord with other recent models, giant clumps of masses M_clump ~ 5×10⁸–10⁹M _☉ form in situ via gravitational instabilities. However, the galactic winds are critical for their subsequent evolution. The giant clumps are short-lived and disrupted by wind-driven mass loss. They do not virialize or migrate to the galaxy centers. These theoretical results are in line with our recent analysis of the resolved stellar light and mass distributions of large samples of 0.5 < z < 2.5 star-forming galaxies, which revealed that bright star-forming clumps generally do not correspond to local peaks in the stellar surface density distribution of galaxies. This implies that they may be rapidly destroyed – plausibly via strong star-formation-driven feedback – unless they migrate to the center within a dynamical timescale (see highlight on "Smoother stellar mass maps").

These results appeared in Genel et al. 2012, ApJ, 745, 11.

SINS/zC-SINF Reveals the Roots of Vigorous Star-Formation-Driven Gas Outflows at z ~ 2

This figure shows the evidence for star-formation-driven outflows from our deep, high-resolution SINFONI+AO observations of z ~ 2 galaxies. Left: Broad and blueshifted wings in the Hα+[NII] line emission profiles are detected at and around the location of several bright star-forming clumps in the z = 2.19 galaxy ZC406690. Insets show the spectra extracted at the position of bright Hα emission peaks along the star-forming ring, marked in the color-composite map of the Hα (green) and rest-frame UV (red) emission. The two maps at the bottom show the distributions of the emission in the narrow component-tracing star-forming sites and of the broad underlying emission-tracing outflowing gas. These observations were the first to provide direct evidence that the origin of the ubiquitous galactic winds long observed on large > 10 kiloparsec scales around distant star-forming galaxies can be traced to extended regions within galaxies and most prominently from the actively star-forming clumps. Top right: The co-averaged spectrum of clumps in five of the SINS/zC-SINF galaxies shows that the broad emission component is also present in fainter clumps, strengthening the evidence from ZC406690, the brightest galaxy of our sample. Bottom right: By co-averaging the integrated spectra of 27 SINS/zC-SINF galaxies that do not host an AGN, we found a strong trend of increasing ionized gas outflow strength (quantifed by the ratio of the flux in the broad and narrow Hα components) with star formation rate surface density, with an apparent threshold at ~ 1 M☉/yr/kiloparsec2, about ten times higher than the wind breakout threshold observed in nearby starburst galaxies. The mass outflow rates inferred for the disks above this threshold are comparable to the star formation rates (and up to several times higher for bright clumps), implying that the outflows can efficiently drive large amounts of gas outside of the galaxies. — This figure shows the evidence for star-formation-driven outflows from our deep, high-resolution SINFONI+AO observations of z ~ 2 galaxies. Left: Broad and blueshifted wings in the Hα+[NII] line emission profiles are detected at and around the location of several bright star-forming clumps in the z = 2.19 galaxy ZC406690. Insets show the spectra extracted at the position of bright Hα emission peaks along the star-forming ring, marked in the color-composite map of the Hα (green) and rest-frame UV (red) emission. The two maps at the bottom show the distributions of the emission in the narrow component-tracing star-forming sites and of the broad underlying emission-tracing outflowing gas. These observations were the first to provide direct evidence that the origin of the ubiquitous galactic winds long observed on large > 10 kiloparsec scales around distant star-forming galaxies can be traced to extended regions within galaxies and most prominently from the actively star-forming clumps. Top right: The co-averaged spectrum of clumps in five of the SINS/zC-SINF galaxies shows that the broad emission component is also present in fainter clumps, strengthening the evidence from ZC406690, the brightest galaxy of our sample. Bottom right: By co-averaging the integrated spectra of 27 SINS/zC-SINF galaxies that do not host an AGN, we found a strong trend of increasing ionized gas outflow strength (quantifed by the ratio of the flux in the broad and narrow Hα components) with star formation rate surface density, with an apparent threshold at ~ 1 M_☉/yr/kiloparsec², about ten times higher than the wind breakout threshold observed in nearby starburst galaxies. The mass outflow rates inferred for the disks above this threshold are comparable to the star formation rates (and up to several times higher for bright clumps), implying that the outflows can efficiently drive large amounts of gas outside of the galaxies.

This figure shows the evidence for star-formation-driven outflows from our deep, high-resolution SINFONI+AO observations of z ~ 2 galaxies. Left: Broad and blueshifted wings in the Hα+[NII] line emission profiles are detected at and around the location of several bright star-forming clumps in the z = 2.19 galaxy ZC406690. Insets show the spectra extracted at the position of bright Hα emission peaks along the star-forming ring, marked in the color-composite map of the Hα (green) and rest-frame UV (red) emission. The two maps at the bottom show the distributions of the emission in the narrow component-tracing star-forming sites and of the broad underlying emission-tracing outflowing gas. These observations were the first to provide direct evidence that the origin of the ubiquitous galactic winds long observed on large > 10 kiloparsec scales around distant star-forming galaxies can be traced to extended regions within galaxies and most prominently from the actively star-forming clumps. Top right: The co-averaged spectrum of clumps in five of the SINS/zC-SINF galaxies shows that the broad emission component is also present in fainter clumps, strengthening the evidence from ZC406690, the brightest galaxy of our sample. Bottom right: By co-averaging the integrated spectra of 27 SINS/zC-SINF galaxies that do not host an AGN, we found a strong trend of increasing ionized gas outflow strength (quantifed by the ratio of the flux in the broad and narrow Hα components) with star formation rate surface density, with an apparent threshold at ~ 1 M_☉/yr/kiloparsec², about ten times higher than the wind breakout threshold observed in nearby starburst galaxies. The mass outflow rates inferred for the disks above this threshold are comparable to the star formation rates (and up to several times higher for bright clumps), implying that the outflows can efficiently drive large amounts of gas outside of the galaxies.

Our newest and deep SINFONI+AO observations of z ∼ 2 star-forming disks allowed us to trace the origin of powerful outflows of ionized gas in non-AGN galaxies. The outflow signature, in the form of a broad FWHM ~ 400 to 500 km/s and blueshifted Hα+[NII] emission component, had been first seen in the co-added integrated spectrum of our initial SINFONI data obtained mostly at seeing-limited resolution (highlighted here). The higher resolution and sensitivity of our new AO-assisted data revealed that the outflows are spatially extended across the galaxies over at least a few kiloparsecs, and most prominent in the immediate vicinity of giant, luminous star-forming clumps. The inferred mass outflow rates from the clumps and the disks are comparable to and even several times the star formation rates, implying that some of the clumps may lose much of their initial mass and dissolve rapidly in the disk before they can migrate to the center of the galaxy. In the galaxy with brightest clumps and highest S/N data, our analysis of line ratio diagnostics ([NII]/Hα and [SII]/Hα) together with photoionization and shock models showed that the emission around the clumps is due to a combination of photoionization from the newly formed massive stars and shocks generated in the outflowing gas component, with 5 to 30% of the emission deriving from shocks.
Among the 27 SINS/zC-SINF non-AGN galaxies observed with SINFONI+AO, we find from co-averaged spectra in bins of global galaxy properties that the inferred gas outflow strength correlates most strongly with the averaged star formation rate surface density, with an apparent threshold for powerful winds around 1 M_☉/yr/kiloparsec². Above this threshold, galaxies with log(M_∗) > 10 have similar or perhaps greater wind mass-loading factors (η = dM_out/SFR) and faster outflow velocities than lower-mass galaxies, suggesting that the majority of outflowing gas at z ∼ 2 may derive from high-mass star-forming galaxies. The threshold at z ~ 2 is an order of magnitude higher than in nearby starbursts that drive galactic-scale winds. In the framework of a simple model where the wind breakout is governed by pressure balance in the disk, the threshold for strong outflows and the mass loading derived from our observations can be explained by the higher ISM pressure in turbulent, gas-rich, and highly star-forming z ~ 2 disks.

These results appeared in a series of three papers by Genzel et al. 2011, ApJ, 733, 101, Newman et al. 2012a , ApJ, 752, 111, and Newman, S.F., et al. 2012b, ApJ, 761, 43.

2011

Galaxy Structure in the Star Formation Rate – Mass Plane from z ~ 2.5 until Today

In parallel to our studies of galaxy kinematics with SINFONI, we analyzed how the structure of galaxies depends on their current star formation rate and amount of assembled stellar mass. Our sample comprised 640,000 galaxies at z ~ 0.1, 130,000 galaxies at z ~ 1, and 36,000 galaxies at z ~ 2. Size and profile measurements for all but the z ~ 0.1 galaxies were based on high-resolution HST imaging, and star formation rates were derived using a Herschel-calibrated ladder of star formation indicators. We found that a correlation between the structure and stellar population of galaxies (i.e., a "Hubble sequence") was already in place as early as z ~ 2.5. At each epoch, the galaxy population can be divided into three classes that coexist over more than an order of magnitude in stellar mass, but differ in star formation activity. Most of the normal star-forming galaxies feature shallow surface brightness profiles indicative of a disk-like nature. At fixed mass, they also tend to have the largest size. More compact and cuspier morphologies are found for quiescent galaxies that already formed the bulk of their stars, and reside below the main sequence of star formation. These results imply that the processes of star formation quenching and bulge formation are closely related. It is tantalizing to speculate that the rare population of starbursting outliers above the main sequence may represent an intermediate evolutionary phase, linking the normal star-forming and quiescent populations. While their star formation is peaking, we are witnessing the rapid build-up of a central cusp that is characteristic of quiescent galaxies. Assuming all starbursting outliers will be quenched, simple duty cycle arguments assign typical timescales ~ 100 Myr for this short-lived phase.

These results appeared in Wuyts et al. 2011a, ApJ, 738, 106, and Wuyts et al. 2011b, ApJ, 742, 96.

Dynamics and Evolution of Giant Star-Forming Clumps in z ~ 2 disks

The images show two examples of z ~ 2 rotating disk galaxies with prominent star-forming clumps: zC406690 at z = 2.19 (left panels) and Q2346-BX482 at z = 2.26 (right panels). For each galaxy, the velocity field, velocity dispersion map, and spatial distribution of star formation map from our SINFONI+AO Hα observations at a resolution of ~ 1 to 2 kiloparsecs are shown in the top row, along with the spatial distribution of the Toomre Q parameter derived from the Hα line emission and kinematic maps. The clumps correspond to minima in the Q maps, with values below unity consistent with their having formed via gravitational instabilities in a turbulent, gas-rich disk. For the first time, the quality of our data allowed us to investigate kinematic signatures of clumps from maps of the velocity residuals, i.e., subtracting the velocities from the best-fitting model disk to the observed velocity field (bottom row). Hα light and residual velocity profiles across the brightest clump in each of the galaxy shown here (arrows overplotted on the bottom row maps indicate where the plotted profiles were extracted) reveal measurable but modest velocity gradients.

The images show two examples of z ~ 2 rotating disk galaxies with prominent star-forming clumps: zC406690 at z = 2.19 (left panels) and Q2346-BX482 at z = 2.26 (right panels). For each galaxy, the velocity field, velocity dispersion map, and spatial distribution of star formation map from our SINFONI+AO Hα observations at a resolution of ~ 1 to 2 kiloparsecs are shown in the top row, along with the spatial distribution of the Toomre Q parameter derived from the Hα line emission and kinematic maps. The clumps correspond to minima in the Q maps, with values below unity consistent with their having formed via gravitational instabilities in a turbulent, gas-rich disk. For the first time, the quality of our data allowed us to investigate kinematic signatures of clumps from maps of the velocity residuals, i.e., subtracting the velocities from the best-fitting model disk to the observed velocity field (bottom row). Hα light and residual velocity profiles across the brightest clump in each of the galaxy shown here (arrows overplotted on the bottom row maps indicate where the plotted profiles were extracted) reveal measurable but modest velocity gradients.

New and deep SINFONI+AO observations of five z ∼ 2 star-forming disks allowed us for the first time to constrain the properties of individual giant star-forming clumps to empirically test scenarios of their formation and evolution. We found that the clumps reside in disk regions where the Toomre Q parameter is below unity, consistent with their being bound and having formed from gravitational instabilities. The clumps leave a modest imprint on the gas kinematics. Velocity gradients across the clumps are 10 to 40 km/s/kiloparsec, similar to the galactic rotation gradients. Given beam smearing and clump sizes, these gradients may be consistent with significant rotational support in typical clumps. The brightest, extreme clumps may not be rotationally supported; either they are not virialized or they are predominantly pressure-supported. The velocity dispersion is elevated and fairly constant across the galaxies, and increases only weakly with star formation surface density. The large velocity dispersions may be driven by the release of gravitational energy, either at the outer disk/accreting streams interface where gas from the halo is infalling onto the disk, and/or by the clump migration within the disk.

These results appeared in Genzel et al. 2011, ApJ, 733, 101.

Pilot HST Near-IR Study: Stellar Properties of Clumps in SINS z ~ 2 Disks

This figure shows two of the SINS z ~ 2 disk galaxies, with evidence suggesting radial variations in the evolutionary stage of clumps. For BX482 (left), a total of seven clumps are identified in the combined HST/NICMOS2 H-band (1.6 μm) and SINFONI+AO Hα maps. The ratio of Hα line flux to rest-optical continuum flux density from the H-band data varies monotonically with the age of a stellar population. The ratios measured for the clumps reveal a trend whereby the clump closest to the center of BX482 is the oldest. For MD41 (right), seven clumps are also identified in the HST/NICMOS2 H-band and HST/ACS I-band (0.8 μm) maps. The observed I–H colors of a stellar population correlate closely with the stellar mass to rest-frame optical light ratio, also an age indicator. The clumps in MD41 tend to be redder at smaller radii, which could be due to older ages. These radial trends provide empirical support for the scenario in which clumps formed in turbulent, gas-rich disks migrate inward, eventually coalescing and contributing to early bulge growth.

This figure shows two of the SINS z ~ 2 disk galaxies, with evidence suggesting radial variations in the evolutionary stage of clumps. For BX482 (left), a total of seven clumps are identified in the combined HST/NICMOS2 H-band (1.6 μm) and SINFONI+AO Hα maps. The ratio of Hα line flux to rest-optical continuum flux density from the H-band data varies monotonically with the age of a stellar population. The ratios measured for the clumps reveal a trend whereby the clump closest to the center of BX482 is the oldest. For MD41 (right), seven clumps are also identified in the HST/NICMOS2 H-band and HST/ACS I-band (0.8 μm) maps. The observed I–H colors of a stellar population correlate closely with the stellar mass to rest-frame optical light ratio, also an age indicator. The clumps in MD41 tend to be redder at smaller radii, which could be due to older ages. These radial trends provide empirical support for the scenario in which clumps formed in turbulent, gas-rich disks migrate inward, eventually coalescing and contributing to early bulge growth.

We studied the stellar properties of kiloparsec-sized clumps identified in the six galaxies observed as part of our pilot program of near-IR (1.6 μm) imaging follow-up with HST of our SINS sample (see the highlight "A Pilot HST Study"). Typically, several clumps are identified in each galaxy, individual clumps contribute a few percent of the galaxy-integrated rest-frame ~ 5,000 Å light, and the total contribution of clump light ranges from around 10 to 25%. The typical clump size and stellar mass are ~ 1 kiloparsec and ~ 10⁹M_☉. These values are within the ranges inferred for clumps identified in rest-UV or Hα line emission in other studies. These properties are consistent with expectations for clump formation through gravitational instabilities in gas-rich, turbulent disks (see highlights "From rings to bulges" and "Dynamics and evolution of clumps"). For two galaxies, the combination of our HST/NICMOS imaging with available SINFONI+AO Hα for one, and HST/ACS rest-UV imaging for the other, at similar kiloparsec-scale resolution reveals trends of higher Hα-equivalent width and redder rest-frame UV-optical colors at smaller galactocentric radius, in contrast to the interclump regions that exhibit little if any radial gradient. This trend can be attributed to older stellar ages of clumps nearer the galaxy center, consistent with the scenario in which massive clumps can migrate inward and contribute to form young bulges in early massive disks.

These results appeared in Förster Schreiber et al. 2011, ApJ, 739, 45.