Bridging the Scales of Star Formation: Embedded Clusters to Galactic Histories

Abstract

Stars are the fundamental tracers of how galaxies form, evolve, and enrich the Universe. They produce much of the visible light we observe from galaxies, synthesize and redistribute heavy elements, and preserve records of the physical conditions under which they formed. In this thesis, I study star formation across a wide range of physical scales, from the emergence of individual young stars from dusty molecular clouds to the recent star formation histories of entire galactic disks. Across these chapters, I use simulations, synthetic observations, and resolved stellar population measurements from nearby galaxies to ask how stars form, when they become observable, and how their present-day properties encode their birth environments. I begin by examining the transition from embedded to exposed star formation using the STARFORGE simulations. By tracking the evolution of embeddedness in simulated star-forming clouds, I find that the transition from embedded to exposed is rapid for individual stars, occurring within 1.3 Myr after each star reaches its maximum mass. Massive stars dominate this transition. They remain highly obscured while accreting, emerge when their feedback overcomes the local inflow, and then drive local gas clearance through pre-supernova feedback. Because these stars dominate the luminosity of young populations, their rapid emergence controls how stellar light escapes from dust. I then move from individual star-forming regions to young stellar clusters by measuring the high-mass stellar initial mass function in Local Group galaxy M33. Using resolved \textit{Hubble Space Telescope} photometry of young clusters, I probabilistically model the optical color-magnitude diagrams of each cluster and infer the distribution of high-mass mass-function slopes. For nine clusters more massive than \(\log(M/M_\odot)=3.6\), I find an ensemble high-mass slope of \(\overline{\Gamma}=1.49\pm0.18\), with no significant intrinsic scatter. This result is consistent with a universal IMF and shows no clear dependence on local star formation rate or galactocentric radius within M33. At larger scales, I use resolved stellar populations to reconstruct the recent star formation history of M31. Using HST optical imaging from the Panchromatic Hubble Andromeda Southern Treasury, I fit color-magnitude diagrams in more than 6500 spatial regions across the southern disk and recover star formation histories over the last \(\sim500\) Myr. Combined with existing measurements from the northern third of the M31 disk, these data provide homogeneous, spatially resolved star formation histories across two-thirds of M31’s star-forming disk. The resulting maps trace the ringed structure of the galaxy and reveal a clear decline in the recent star formation rate, with a pronounced drop over the last \(\sim40\) Myr. The decline is strongest in the star-forming rings and in regions near M32, suggesting that M31 is in the late stages of a longer-term wind-down from a more active state. Motivated by this result, I finally examine how reliably FUV luminosity traces recent star formation when the star formation history is evolving. FUV luminosities are commonly interpreted as star formation rates averaged over the last \(\sim100\) Myr, but that calibration assumes a constant recent star formation history. By modeling a range of idealized histories, I show that many different recent star formation histories can produce the same present-day FUV luminosity. When the star formation rate is not constant, the FUV becomes increasingly uncertain as a constraint on the true average rate beyond \(\sim35\) Myr. Across rising and declining histories, averaging timescales of \(\sim25\)--40 Myr provide better agreement between the FUV-inferred and true input star formation rates than the traditional 100 Myr average. Together, these results from the four chapters ofd this thesis connect the physics of star formation to the stellar populations that survive as observable records.