My Research


What is the assembly history of the Milky Way? How does its evolution compare to smaller dwarf galaxies? What are the relative yields of different elements from stars and supernovae?

I am broadly interesting in galaxy evolution and the origin of the elements. Galactic chemical evolution (GCE) bridges the gap between nuclear physics and astrophysics by combining galactic processes like star formation with nucleosynthetic yields from stellar evolution models to estimate enrichment rates and elemental abundances in the interstellar medium (ISM). By studying the chemical compositions of stars, we can understand both the evolutionary histories of their host galaxies and the stellar evolution processes that shaped their nucleosynthetic yields.

My work draws heavily on numerical and statistical modeling applied to large datasets from modern spectroscopic surveys like APOGEE, SDSS-V's new Milky Way Mapper program, and the H3 Survey.

Dwarf Galaxies

One-zone model fit to Gaia-Sausage Enceladus
The simplest models of galactic chemical evolution (GCE) are called "one-zone" models, which assume instantaneous mixing in the interstellar medium (ISM). Due to their small sizes, turbulent velocities, and inefficient star formation, this approximation likely applies best to dwarf galaxies.

This figure shows a fit of one of these models to the Gaia-Sausage Enceladus (GSE), a disrupted dwarf galaxy in the inner stalo halo of the Milky Way. GSE exhibits the behavior of a "classic" chemical evolution model, where the [α/Fe] ratio decreases with increasing [Fe/H]. This work follows a novel, statistically rigorous star-by-star fitting method, the details of which can be found in Johnson et al. (2022b).
accretion-driven starburst
Dwarf galaxies often have bursty star formation histories. Isolating this effect, Johnson & Weinberg (2020) quantify the effects of simple starbursts in one-zone GCE models. In any starburst scenario, the principle observational signature is a surplus of high [α/Fe] stars, as in this figure. The details contain information on the timing and strength of the burst, and if the burst's decay is accompanied by quick drops in the star formation efficiency, low [α/Fe] stars can form as well. Here α refers to an "alpha element," specifically O or Mg, whose enrichment is dominated by metallicity-independent yields from massive stars.
accretion-driven starburst

The Milky Way

GCE evolutionary tracks incorporating stellar migration
In the Milky Way, processes such as stellar migration and radial gas flows have prompted the development of "multi-zone" GCE models stitching together many one-zone models to describe the abundances in different Galactic regions. Johnson et al. (2021) presents one of these models combining inside-out galaxy growth with stellar migration. This empirically motivated combination leads to a natural explanation of many details of the Galactic abundance structure. The assembly history of the Milky Way is a topic of debate, so I continually investigate variations of this model.

Versatile Integrator for Chemical Evolution

What it is

VICE is an open-source GCE modeling software that I created. Many evolutionary parameters, such as the galaxy's star formation history, are arbitrary, user-defined functions. Best described as a "one size fits all" GCE software, this implementation allows users to explore highly complex parameterizations. VICE also provides a variety of nucleosynthetic yield tables from stellar evolution studies to explore.

Why to use it

VICE is a python package written in C, allowing it to enjoy both the powerful computing speeds of a compiled language and the ease of use of python. In keeping with its mantra of flexibility, users may construct their own expressions for nucleosynthetic yields independent of stellar evolution models. VICE users also enjoy excellent support provided by a team of developers!

The Origin of the Elements

GCE predicted nitrogen and oxygen abundances
Despite how important they are for GCE models, nucleosynthetic yields are poorly understood. Nitrogen is one particularly difficult element to accurately model, because it is highly sensitive to uncertain details of stellar evolution models, such as rotational mixing. In Johnson et al. (2022a), I used my GCE models of the Milky Way to constrain nitrogen yields from stellar populations, finding that their metallicity dependence must be approximately linear with Z in order to explain the correlation between nitrogen and oxygen abundances (see Figure). I am collaborating with OSU students on addressing similar questions for carbon and helium, and plan to investigate additional elements in the future.

These projects are also closely related to collaborative work I've done with Ryan Cooke on interpreting helium isotope ratios in the ISM and with Emily Griffith and David Weinberg on the two-process model deducing relative yields from the separation between the chemical thin and thick disks.

Other Work

Metallicity effects of the specific SN Ia rate
If all galaxies were the same, then each galaxy would produce similar numbers of supernovae. Instead, low stellar mass galaxies have higher specific Type Ia supernova (SN Ia) rates than more massive galaxies (scaling as ~M-0.3, shallow black dashed line in the figure). Low mass galaxies are empirically known to host lower metallicity stellar populations, and the close binary fraction increases toward low metal abundances as well. In Johnson, Kochanek & Stanek (2022), we found that this leads to a natural explanation of the observations, suggesting that the elevated SN Ia rates arise due to a higher binary fraction as a consequence of the low metallicities of dwarf galaxies.
Metallicity effects of the specific SN Ia rate
Halo spin assembly bias
As an undergraduate, I studied the origins of halo spin assembly bias, which refers to the result from cosmological N-body simulations that high spin dark matter halos at fixed mass tend to be found in denser environments than their low spin counterparts. In collaboration with Ari Maller and Andreas Berlind, who was a faculty member at Vanderbilt University at the time, we found that similar mass halos in close proximity to one another tend to have slightly higher spin due to perturbations from the local tidal field (see Figure). In Johnson et al. (2019), we demonstrate that this "twin bias" accounts for the majority of the effect. Though the correlation between halo spin and twin distance is subtle, twin halos have a substantial effect on statistical measurements of spatial clustering.