When Did Elliptical Galaxies Form?

Adam Stanford (LLNL), Peter Eisenhardt (JPL), and Mark Dickinson (JHU) are using NOAO near-IR imagers to deduce the formation epoch of elliptical galaxies. Their goal is to look for scatter in the optical-IR colors of high-redshift ellipticals galaxies that would indicate relatively recent star-formation at the corresponding lookback time. Their observations suggest that most of the stars in luminous elliptical galaxies had already formed when the universe was less than 25% of its current age.

The standard elliptical galaxy formation scenario begins with a single burst of star formation at high redshift; ellipticals would form the vast majority of their stellar mass during this initial episode. Several observations suggest that present-epoch ellipticals do in fact descend from such a coeval population formed at z > 3. Notably, little scatter at a given galaxy luminosity is seen in the UVK colors of present-epoch E/S0s such as those in Coma (Bower, Lucey, and Ellis 1992). Restframe broadband colors spanning the 4000Å break are sensitive to recent star formation---even small starbursts greatly increase the U band flux, and thus the scatter in colors like U-H. However, critical application of this test requires the colors to be observed at high-redshift. For example, an alternative galaxy formation scenario developed in the context of the cold dark matter (CDM) model suggests that massive galaxies form later, at z 2, from the gradual merging of smaller galaxies. The small color scatter seen in nearby ellipticals can also be explained by CDM because, even though this model entails more recent star formation compared to the high z scenario, sufficient time elapses from z ~ 2 to z ~ 0 for starburst-induced color variations to damp out. Accordingly, the CDM model predicts an increase in the color scatter at higher redshifts. Until recently, the scatter in the color-mag relation of early-type galaxies in clusters was well-determined only at z ~ 0, which allowed both galaxy formation theories a wide range of possible star formation histories.

Caption:The bottom panel shows median colors of cluster E/S0s vs redshift. The colors are zeropointed to a no-evolution prediction based on Coma E/S0 photometry. Errorbars include systematics; they do not represent scatter. The top panel shows intrinsic color scatter against redshift for the early-type galaxies in a subsample of z < 1 clusters. The dashed line is the intrinsic scatter for a sample of Coma E/S0s.

Using NOAO telescopes over the last four years, Stanford, Eisenhardt, and Dickinson have been conducting a large scale optical-IR imaging survey of galaxy clusters from Coma to z ~ 0.9. Using WFPC2 images currently available from the HST Archive of 16 of the 40 clusters in their sample, early-type galaxies were selected by their morphologies so that their color distribution could be measured with minimal contamination. The resulting subsample contains ~480 E/S0s, down to 2 mag below a no-evolution K. The optical-K colors of the E/S0s become bluer with redshift, relative to the present epoch as shown in the bottom panel of the figure. The measured color changes are consistent with the predictions of a passive-evolution elliptical from the Bruzual and Charlot (solar metallicity version) models for the formation-redshift, , , ---that is, the only major thing that is happening is the slow aging of the stellar population after z. Moreover, after accounting for observational error, the intrinsic scatter in the optical-IR colors of the E/S0s was found to be small, and to be roughly constant with redshift, as shown in the top panel of the figure. Following the method outlined in Bower, Lucey, and Ellis (1992), a limit on the formation redshift may be obtained from the color scatter. Using GHO 1603+4313, their highest redshift cluster at 0.895, Stanford et al. find for a reasonable degree of coherence among the cluster E/S0s during their formation epoch. This formation redshift is in closer agreement with the single-burst scenario than with the CDM model. Stanford it et al. plan to continue to study the properties of early-type galaxies in high redshift clusters by using the sample at z > 1, which Dickinson et al. have been assembling (see NOAO Newsletter No. 37).