2009ApJ...699..254H

Data from the Hubble Space Telescope taken with the Advanced Camera for Surveys/WFC camera have been used to investigate the globular cluster (GC) populations around six giant elliptical galaxies that are ∼40 Mpc distant. From these six fields, imaged in B and I, a total of more than 15,000 candidate GCs have been measured, of which 8000 or more are high-probability GCs. The data reach a limiting magnitude near M_I≃ -8, about 0.4 mag fainter than the GC luminosity function turnover point, and thus thoroughly cover the bright half of the GC population. Most of the individual GCs on these images are marginally resolved nonstellar objects, so King-model profiles convolved with the stellar point-spread functions are used to measure their individual total magnitudes, colors, and linear effective radii. The classic bimodal form of the GC color-magnitude distribution shows up unambiguously in all the galaxies, allowing an accurate definition of the mean colors along each of the two sequences as a function of magnitude (the mass/metallicity relation or MMR). The blue, metal-poor cluster sequence shows a clearly defined but nonlinear MMR: in this particular photometric data set the mean GC color changes smoothly from a near-vertical sequence at low luminosity (M_I≳ -9.5) to an increasingly redward slope at higher luminosity. By contrast, the red, metal-rich sequence shows little trace of an MMR and is nearly vertical at all luminosities. The form and slope of the MMR along either sequence do not depend strongly on either cluster size r_h or galactocentric distance R_gc. All the observed features of the present data agree with the interpretation that the MMR is created primarily by GC self-enrichment, along the lines of the quantitative model of Bailin & Harris. During the protocluster formation stage, the more massive GCs are better able to hold back the enriched products of the earliest supernovae and to seed the lower-mass stars still in formation. The "threshold" mass at which this effect should become noticeable is near 1 million Solar masses, which is closely consistent with the transition region that is seen in the data. More generally, the data favor models in which the star formation efficiency in a protocluster is roughly independent of mass, and in which the gas retention efficiency is a strong function of mass. Correlation of the median scale sizes r_h of the GCs with other parameters shows that the metal-poor clusters are consistently 17% larger than those of the metal-rich clusters, and that this difference holds at all galactocentric distances and luminosities. At the same time, cluster size scales with halo location as r_h∼ R ^0.11_gc, indicating that both metallicity and the external tidal environment play roles in determining the scale size of a given cluster. Lastly, both the red and blue GC components show metallicity gradients with galactocentric distance that are shallow but real: heavy-element abundance scales as Z ∼ R ^–0.1_gc for both types.