Turbulent Deaths

Title: The Three Dimensional Evolution to Core Collapse of A Massive Star
Authors: S. M. Couch, E. Chatzopoulos, W. D. Arnett, F. X. Timmes
First Author’s Institution: TAPIR, Walter Burke Institute for Theoretical Physics, California Institute of Technology, Pasadena, CA
Status: Submitted to ApJ Letters


There are hordes of them out there.  Giant behemoths that masquerade as massive stars, but that never birthed a single radiant photon nor fused a pair of hydrogen nuclei.   All of them are found on Earth.  Instead of atoms, they’re built of up strings of 0’s and 1’s and live in computers of all shapes and sizes across the world (though they’d prefer the roomier accommodations of supercomputers, if you ask them). Inspired by their real, yet enigmatic counterparts in the physical universe, we first brought them to life in simple, spherically symmetric, one-dimensional form.  We quickly bestowed them full three-dimensional complexity and an increasingly comprehensive set of input physics as soon as our computers possessed the computational brawn to handle them, feathery convective plumes and other instabilities, mottled compositional complexions, freewheeling invisible neutrinos, and all.

All for one goal: to understand how they die. We’ve observed their real counterparts supernovae over and over and over again, at a rate of about one every five seconds, bursting in galaxies near and far, their spectacular showers of photons (sometimes rivaling that of an entire galaxy) often traveling cosmic distances to reach our army of telescopes on Earth.  We’ve sought to replicate these supernovae in our virtual massive stars, giving them encouraging nudges by injecting extra boosts of energy-imparting, explosion-inducing neutrinos, giving them a bit of a spin, tweaking how their mass is distributed, sending in sound waves, draping them with magnetic fields.

And yet.  Despite the care with which we’ve crafted them, our virtual massive stars almost always refuse to explode.


What might we be missing in our theories of massive stellar death? It’s a question we’ve been asking for decades. Instead of focusing on the properties of the stellar cores that collapse and usher in their deaths, the authors of today’s paper instead turned to consider the life of the star preceding. They were motivated particularly by hints that the silicon-burning shell surrounding the pre-collapse core could be violently turbulent, stirred by convective motions in the shell. The authors thus concocted a new star, one 15 times the mass of our Sun. They harnessed the power of MESA, a special-purpose code built specifically for modeling the life of stars in 1D, from their early lives burning hydrogen, then helium, carbon, all the way to silicon and the formation of an iron core, about three minutes shy of core collapse. In order to focus on the effects of a convection in a silicon-burning shell, they stripped their stars of any complexifying qualities: no rotation, no magnetic fields.

Figure 1. Development of convection pre-collapse.

Figure 1. Development of convection pre-collapse.  The ring-like structure seen here is the silicon-burning shell within the star the authors modeled.  A map of the amount of silicon (by mass, top left) reveals intricate swirls and eddies caused by convection in the shell about 2.5 minutes before core collapse.  Bottom left: a map of where burning occurs.  Bottom right: the distribution of iron (by mass), mostly found at the center of the star.  Top right: the speed of the gas (redder is faster).  For a video showing how the convection develops, see the movie the authors made here.

At this point, however, their star possessed no convection, which cannot develop in 1D. Thus the authors turned to FLASH, a powerful hydrodynamics code that can follow the evolution of the complex gas motions that give rise to convection in stars. And this time, they let the star evolve in 3D. At the end of this, they had a star with a fully convective silicon-burning shell (see Figure 1), replete with characteristic convective plumes—spectacular ones that spanned the entire width of the silicon-burning shell and churned at velocities of several hundreds of kilometers a second, whirling around an 1.3 solar mass iron core on the verge of collapse.

And then, of course, came the collapse. The authors exploded two stars, twin stars, identical in every way except that one lived in one dimension and was thus spherically symmetric, while the other lived in three (though because of the computational complexity they modeled only an octant of the star) and thus retained its convectively-stirred, complex 3D structures. To help the stars explode, they were given identical shots of extra energy in the form of neutrino heating, then let go. And go they did—and differently in some key ways. Their cores initially evolved much in the same way: they collapsed, rebounded, giving birth to a shock, both of which successfully continued to grow. When the shock reached the silicon-burning shell, substantial differences began to show: the 3D convective star’s shock grew more rapidly than its 1D twin, and had a larger explosion energy. Though the authors did not evolve the collapse long enough to determine whether or not the star eventually exploded, these were promising signs that an explosion could be achieved more readily.

So does this mean that we now have it—the secret to the deaths of massive stars? Not quite. Many assumptions and simplifications—the initial 1D models, the 3D octant of the star, to name a few—were made. But while these new models were necessarily contrived, given the limits of today’s computational brawn, they are still an instructive demonstration that the turbulent environments in which the cores of massive stars breathe their last can affect how the rest of the star’s death plays out.

About Stacy Kim

I was a former graduate student in The Ohio State University's Department of Astronomy. On a day-to-day basis, you could typically have found me attempting to smash clusters of galaxies together inside big supercomputers to see if cluster mergers are good testbeds for dark matter collisionality. As an undergraduate at Caltech, I spent a few years chasing photons where planets are thought to form (or, as they say, performing Monte Carlo radiative transfer calculations of protoplanetary disks) at NASA's Jet Propulsion Laboratory. When I wasn't sitting in front of a computer trying to translate cosmic thoughts into pithy lines of code, you could often find me in the kitchen or on the walls of a climbing gym.

Discover more from astrobites

Subscribe to get the latest posts to your email.


  1. Would the availability of better computing power and thus the ability to simulate more than just an octant of the star be more revealing? I imagine that this will also allow for simulating beyond the point where they ended here if computation power was the limiting factor.

  2. Interesting how the symmetry of a sphere is not very necessary in causing these stellar deaths, as shown by the octant of sphere. Unless the programmers did some kind of trick to involve spherical symmetry but omitted the full computation?

  3. It really is amazing how complex computer simulations have become. I can’t imagine this is a simple model to simulate, having done some basic statistical simulations myself.

  4. Very nice post! One question: what is the difference between having a star collapsing and having it explode?

  5. The style of your writing is fantastic, I really enjoyed the article! I was wondering if you could tell me how much actual data is input into these simulations or compared with predictions? Essentially, where is the balance between theory and observations of stellar deaths?

  6. Could you clarify the exact purpose of running the 1D simulation here in addition to the 3D one? I thought it was meant to simplify the computing, but if the authors run a 3D simulation too then what are we getting from the 1D simulation that we don’t get from that?

  7. Some of the inputs seemed a little arbitrary- where did the extra energy from neutrinos that the scientists inputted come from? Is that just a way to give the star extra energy in the simulations or is that something that occurs in nature?


Leave a Reply