SCYON Abstract

Received on December 10 2012

Evidence for Two Distinct Stellar Initial Mass Functions

AuthorsDennis Zaritsky, Janet E. Colucci, Peter M. Pessev, Rebecca A., Bernstein, and Rupali Chandar
AffiliationUniversity of Arizona, UCO/Lick Observatory, Gemini South Observatory, UCO/Lick Observatory, and University of Toledo
Accepted byAstrophysical Journal


We present velocity dispersion measurements of 20 Local Group stellar clusters (7 < log(age [yr]) < 10.2) from integrated light spectra and examine the evolution of the stellar mass-to-light ratio, Upsilon*. We find that the clusters deviate from the evolutionary tracks corresponding to simple stellar populations drawn from standard stellar initial mass functions (IMFs). The nature of this failure, in which Upsilon* is at first underestimated and then overestimated with age, invalidates potential simple solutions involving a rescaling of either the measured masses or modeled luminosities. A range of possible shortcomings in the straightforward interpretation of the data, including subtleties arising from cluster dynamical evolution on the present-day stellar mass functions and from stellar binarity on the measured velocity dispersions, do not materially affect this conclusion given the current understanding of those effects. Independent of further conjectures regarding the origin of this problem, this result highlights a basic failing of our understanding of the integrated stellar populations of these systems. We propose the existence of two distinct IMFs, one primarily, but not exclusively, valid for older, metal-poor clusters and the other for primarily, but not exclusively, younger, metal-rich clusters. The young (log(age [yr]) < 9.5) clusters are well described by a bottom-heavy IMF, such as a Salpeter IMF, while the older clusters are better described by a top-heavy IMF, such as a light-weighted Kroupa IMF, although neither of these specific forms is a unique solution. The sample is small, with the findings currently depending on the results for four key clusters, but doubling the sample is within reach.