Theoretical predictions for IMF diagnostics in UV spectroscopy of star clusters

TL;DR

This paper investigates UV spectroscopy combined with photometry as a method to determine the IMF slope in unresolved star clusters, extending a spectral synthesis code and applying Bayesian inference to improve parameter recovery.

Contribution

It extends the slug code with UV spectral synthesis and equivalent width calculations, and demonstrates improved IMF slope estimation using combined UV spectral features and photometry.

Findings

Bayesian inference improves IMF slope recovery by 32% with UV data.

Including UV spectral features enhances IMF slope constraints.

Sensitivity to modest equivalent widths significantly improves parameter estimation.

Abstract

We explore the possibility of using UV spectroscopy in combination with broad-band photometry as diagnostic tools for understanding the shape of the Initial Mass Function (IMF) in unresolved stellar populations. Building on our previous work, we extend the Stochastically Lighting Up Galaxies code (slug) to include a high-resolution UV spectral synthesiser and equivalent width calculation capabilities. We first gain a qualitative understanding of how UV spectral features behave as the parameters that define a star cluster in slug (mass, age, extinction, and IMF slope alpha) are changed. We then exploit Bayesian inference techniques to recover the alpha values for clusters simulated with slug, using mock observations of these clusters comprised of broad-band photometry and equivalent width measurements of a selection of UV spectral features. We find some improvement when compared to attempts using broad-band photometry alone (with the interquartile range of the alpha posterior PDF shrinking by 32%), although we still do not yet fully break the known degeneracy between the cluster mass and alpha. Finally, we make predictions about how effective real observations will be by quantifying our ability to constrain alpha as a function of limiting equivalent width. We find that observations sensitive to a modest equivalent width of 9 A are sufficient to improve the recovery of the IMF slope parameter by 32% (interquartile range of posterior PDF median residuals), moving to 39% when we include all the significant spectral features in the wavelength range 900-3000 A.