We present new evidence for mature stellar populations with ages >100 Myr in massive galaxies (M stellar > 10 10 M ) seen at a time when the Universe was less than 1 Gyr old. We analyse the prominent detections of two z ≈ 6 star-forming galaxies (SBM03#1 and #3) made at wavelengths corresponding to the rest-frame optical using the Infrared Array Camera camera onboard the Spitzer Space Telescope. We had previously identified these galaxies in Hubble Space Telescope (HST) Advanced Camera for Surveys (ACS) / Great Observatories Origins Deep Survey (GOODS) images of Chandra Deep Field South through the 'idrop' Lyman-break technique, and subsequently confirmed the identification spectroscopically with the Keck telescope. The new Spitzer photometry reveals significant Balmer/ 4000-Å discontinuities, indicative of dominant stellar populations with ages >100 Myr. Fitting a range of population synthesis models (for normal initial mass functions) to the HST/Spitzer photometry yields ages of 250-650 Myr and implied formation redshifts z f ≈ 7.5-13.5 in presently-accepted world models. Remarkably, our sources have best-fitting stellar masses of 1.3-3.8 × 10 10 M (95 per cent confidence) assuming a Salpeter IMF. This indicates that at least some galaxies with stellar masses >20 per cent of those of a present-day L * galaxy had already assembled within the first Gyr after the Big Bang. We also deduce that the past average star formation rate must be comparable to the current observed rate (SFR UV ∼ 5-30 M yr −1 ), suggesting that there may have been more vigorous episodes of star formation in such systems at higher redshifts. Although a small sample, limited primarily by Spitzer's detection efficiency, our result lends support to the hypothesis advocated in our earlier analyses of the Ultra Deep Field and GOODS HST/ACS data. The presence of established systems at z ≈ 6 suggests that long-lived sources at earlier epochs (z > 7) played a key role in reionizing the Universe.

We measure the ages, stellar masses, and star formation histories of z∼ 6 galaxies, observed within 1 Gyr of the big bang. We use imaging from the Hubble Space Telescope (HST) and the Spitzer Space Telescope from the public ‘Great Observatories Origins Deep Survey’ (GOODS), coupled with ground‐based near‐infrared imaging, to measure their spectral energy distributions (SEDs) from 0.8–5 μm, spanning the rest‐frame ultraviolet (UV) and optical. From our sample of ≈50 ‘i′‐drop’ Lyman‐break star‐forming galaxies in GOODS‐South with z′AB < 27, we focus on ≈30 with reliable photometric or spectroscopic redshifts. Half of these are confused with foreground sources at Spitzer resolution, but from the 16 with clean photometry we find that a surprisingly large fraction (40 per cent) have evidence for substantial Balmer/4000‐Å spectral breaks. This indicates the presence of old underlying stellar populations that dominate the stellar masses. For these objects, we find ages of ∼200–700 Myr, implying formation redshifts of 7 ≤zf≤ 18, and large stellar masses in the range ∼1–3 × 1010 M⊙. Analysis of seven i′‐drops that are undetected at 3.6 μm indicates that these are younger, considerably less massive systems. We calculate that emission line contamination should not severely affect our photometry or derived results. Using SED fits out to 8 μm, we find little evidence for substantial intrinsic dust reddening in our sources. We use our individual galaxy results to obtain an estimate of the global stellar mass density at z∼ 6. Correcting for incompleteness in our sample, we find the z∼ 6 comoving stellar mass density to be 2.5 × 106 M⊙ Mpc−3. This is a lower limit, as post‐starburst and dust‐obscured objects, and also galaxies below our selection thresholds, are not accounted for. From our results, we are able to explore the star formation histories of our selected galaxies, and we suggest that the past global star formation rate may have been much higher than that observed at the z∼ 6 epoch. The associated UV flux we infer at z > 7 could have played a major role in reionizing the Universe.

We present a new measurement of the integrated stellar mass per comoving volume at redshift 5 determined via spectral energy fitting drawn from a sample of 214 photometrically selected galaxies with z 0 850LP < 26:5 in the southern GOODS field. Following recent procedures introduced by Eyles et al., we estimate stellar masses for various subsamples for which reliable and unconfused Spitzer IRAC detections are available. A spectroscopic sample of 14 of the most luminous sources with z ¼ 4:92 provides a firm lower limit to the stellar mass density of 1 ; 10 6 M Mpc À3. Several galaxies in this subsample have masses of order 10 11 M , implying that significant earlier activity occurred in massive systems. We then consider a larger sample whose photometric redshifts in the publicly available GOODS-MUSIC catalog lie in the range 4:4 < z < 5:6. Before adopting the GOODS-MUSIC photometric redshifts, we check the accuracy of their photometry and explore the possibility of contamination by low-z galaxies and low-mass stars. After excising probable stellar contaminants and using the z 0 850LP À J color to exclude any remaining foreground red galaxies, we conclude that 196 sources are likely to be at z ' 5. The implied mass density from the unconfused IRAC fraction of this sample, scaled to the total available, is 6 ; 10 6 M Mpc À3. We discuss the uncertainties, as well as the likelihood that we have underestimated the true mass density. By including fainter and quiescent sources, the total integrated density could be as high as 1 ; 10 7 M Mpc À3 . Even accounting for 25% cosmic variance within a single GOODS field, such a high mass density only 1.2 Gyr after the big bang has interesting consequences for the implied past average star formation during the period when cosmic reionization is now thought to have taken place. Using the currently available ( but highly uncertain) rate of decline in the star formation history over 5 < z < 10, a better fit is obtained for the assembled mass at z ' 5 if we admit significant dust extinction at early times or extend the luminosity function to very faint limits. An interesting consequence of the latter possibility is an abundant population of low-luminosity sources just beyond the detection limits of current surveys. As mass density estimates improve at z ' 5Y6, our method is likely to provide one of the tightest constraints on the question of whether star-forming sources were responsible for reionizing the universe.

Abstract.We discuss the selection of star-forming galaxies at z 6 through the Lyman-break technique. Spitzer imaging implies many of these contain older stellar populations (> 200 Myr) which produce detectable Balmer breaks. The ages and stellar masses (∼ 10 10 M ) imply that the star formation rate density at earlier epochs may have been significantly higher than at z 6, and might have played a key role in re-ionizing the universe.