Title: Molecular gas and star formation in nearby starburst galaxy mergers
Authors: Hao He, Conner Bottrell, Christine Wilson, Jorge Moreno, Blakesley Burkhart, Christopher C. Hayward, Lars Hernquist, and Angela Twum
First Author’s Institution: McMaster University, Canada
Status: Submitted to ApJ Letters [open access]

Even though there are billions of known stars in the universe, we still don’t fully understand how they’re formed. We know the basics: stars form inside giant nurseries made of the coldest, densest phase of hydrogen gas found in galaxies (“molecular gas”). These clouds of molecular gas get so dense that their gravity becomes stronger than the random, turbulent motions of the gas inside the cloud, and the gas collapses into stars. This theory, however, predicts that stars should be forming at a much higher rate than is actually being observed, and we don’t fully understand the reasons for this discrepancy. In today’s paper, the authors investigate one possible explanation: mergers between galaxies, such as the one shown in Figure 1, ripping the molecular gas apart before it has the chance to form stars.
Not actually that Giant
The specifics of star formation are still so unclear to us because of a problem with scale. The molecular gas that makes new stars is usually clumped into clouds, known as Giant Molecular Clouds, or GMCs. Individual GMCs tend to be 5-200 parsecs in size (for context, the distance between earth and sun is 0.00001 parsec), but in order to understand how they’re affected by a galaxy’s evolution, you would need to study the whole galaxy–around 20 kiloparsecs, or a thousand-to-one size ratio. The problem is worsened if you want to simulate whole sections of the universe (megaparsec scales, or a million-to-one ratio), to capture realistic interactions between galaxies. It’s the ratio between these sizes that’s the problem–modern galaxy simulations can resolve individual gas clouds, or cover massive areas of the universe, but they can’t usually do both.
FIRE-ing off some simulations
The authors of today’s paper don’t try to tackle the million-to-one scale problem. Instead, they go for two-thousand-to-one, to see if they can realistically simulate how GMCs are affected when two galaxies smash together. They do this with idealized simulations: simulations in which the user selects the exact individual conditions they want their galaxy or galaxy merger to have, and run the simulations only from that point. This is in contrast to cosmological simulations, where the user tries to let galaxies evolve naturally out of a simulated universe.
The authors simulate two disk galaxies, running a simulation where the galaxies merge and (as a control) a simulation where they evolve without merging. In both cases, they use the FIRE-2 simulation suite. This main algorithm simulates the physics acting on the gas in the galaxy to decide how hot and dense each region should be at each moment in time. They set the resolution of the simulation to around 40 pc, which is small enough that medium-to-big GMCs should be resolved. Once the main simulations are done, the authors run secondary algorithms to convert them into a format that matches telescope observations. The result is a very cool video that you can watch here.
Checking Their Work
Since these authors are trying to simulate something that’s notoriously tricky, how do they know it worked? They answer that question by comparing the simulation results to spatially-resolved observations of molecular gas in actual galaxies, both merging and alone. The authors compare their simulations to two well-known mergers (the Antennae and NGC 3256) and to a sample of 70 isolated galaxies. All these observations have resolutions of around 100 pc, which is a similar resolution to the simulations.
So, how did they do?
Pretty well! The authors quantify their results in terms of a well-known observational relationship between the surface density, Σmol, of molecular gas in a galaxy and the velocity dispersion σv of that gas. Very simply, this relationship says that if there’s more gas, the forces acting on that gas are more extreme and it moves around more. Both external forces and internal chemistry affect the individual parameters, but in real galaxies the balance stays fairly consistent (see Figure 2), whether they’re merging or isolated. This means it works well as a metric for the quality of the simulations.

The authors find that their simulations of individual galaxies (Figure 2, left panel) match observations extremely well, but the numbers for the merger simulations (Figure 2, right panel) fall a little above the relationship. This might be due to some of the initial conditions they used when setting up their simulations. They note that the simulated galaxy’s star formation rate increases in response to more intense molecular gas conditions exactly as would be expected, so the star-formation physics in the simulation seem spot-on.
What’s next?
Galaxy mergers are complicated beasts, and one simulated merger definitely isn’t enough to test the simulation strategy fully or learn much about how mergers affect molecular gas. That makes the next step easy, though: run more simulations! The authors have plans to tweak the initial conditions of their simulated mergers to better match observations and to collect a bigger sample of observed mergers for comparison. If the effects of galaxy mergers on GMCs can be accurately simulated, it will bring us one step closer to understanding the complex processes governing star formation.
Astrobite edited by Samantha Wong and Viraj Karambelkar
Featured image credit: (Adapted from) Hubble Legacy Archive, ESA, NASA; Processing – Martin Pugh