# Variations in $\alpha$-element ratios trace the chemical evolution of the disk

## TL;DR

This study investigates the [Mg/Si] abundance ratio in stars to uncover additional details about the Milky Way's chemical evolution, revealing subtle correlations with stellar properties and highlighting limitations in current models.

## Contribution

It introduces the [Mg/Si] ratio as a new, detailed chemical tracer and demonstrates its potential to provide insights beyond traditional alpha-element measurements.

## Key findings

- [Mg/Si] correlates with alpha-enhancement, age, and galactic location.
- Current chemical evolution models cannot fully reproduce observed [Mg/Si] variations.
- Data-driven approaches are needed to improve nucleosynthetic yield predictions.

## Abstract

It is well established that the chemical structure of the Milky Way exhibits a bimodality with respect to the $\alpha$-enhancement of stars at a given [Fe/H]. This has been studied largely based on a bulk $\alpha$ abundance, computed as a summary of several individual $\alpha$-elements. Inspired by the expected subtle differences in their nucleosynthetic origins, here we probe the higher level of granularity encoded in the inter-family [Mg/Si] abundance ratio. Using a large sample of stars with APOGEE abundance measurements, we first demonstrate that there is additional information in this ratio beyond what is already apparent in [$\alpha$/Fe] and [Fe/H] alone. We then consider Gaia astrometry and stellar age estimates to empirically characterize the relationships between [Mg/Si] and various stellar properties. We find small but significant trends between this ratio and $\alpha$-enhancement, age, [Fe/H], location in the Galaxy, and orbital actions. To connect these observed [Mg/Si] variations to a physical origin, we attempt to predict the Mg and Si abundances of stars with the galactic chemical evolution model Chempy. We find that we are unable to reproduce abundances for the stars that we fit, which highlights tensions between the yield tables, the chemical evolution model, and the data. We conclude that a more data-driven approach to nucleosynthetic yield tables and chemical evolution modeling is necessary to maximize insights from large spectroscopic surveys.