Cosmic Ancestry of the first Neutron Star Merger
A summary of the Nature Astronomy publication “End-to-end study of the home and genealogy of the first binary neutron star merger” aimed at science enthusiasts.
The last 3 years of my work have been focused on tracking down the genealogy tree of the first observed Neutron Star merger. These events are crucial in creating some of the elements that make the world as we know it - such as gold and rare metals needed for our technology - since they were not made during the Big Bang. The methods I have laid out will be used in the future as more mergers and other types of explosions are uncovered so we can better understand where we come from. But what is stellar genealogy, and what does it tell us about the parents of the first kilonova explosion? (If you just want the conclusions see bullet points at the end)
What is “Stellar Genealogy”?
Drawing the family tree of a cosmic explosion is not as easy as typing your parents name into ancestry.com, but it is one of the most exciting challenges in modern astronomy. I call this type of work “Stellar genealogy”. The genealogy of a star system will look a little different from your own, but it starts in the same place - we all have “parents” after all. In this case the “parent system” are the newborn stars that started burning hydrogen in their core millions or billions of years ago. Most often in the work that I do, this system contains two stars in orbit around each other. We call this a binary system and most (if not all!) of the stars that die as supernova explosions and create neutron stars, are born with one companion at least. If you’re going to spend millions of years floating in space you might as well not do it alone.
Our parent star system is defined by three main characteristics: how heavy star 1 is, how heavy star 2 is, and how close they are to each other. These three quantities alone basically determine the life story of our system and what key events might occur. Because yes, binary star systems can have very eventful lives indeed. As a star becomes old and starts burning hydrogen outside of its core, it will expand to become a Red (Super) Giant that can easily reach hundreds of times the size of our Sun! If the binary is in a tight orbit, the outer layers of our giant star can be pushed far enough towards the companion that its gravitational pull becomes dominant. In other words it will start hoovering up the envelope of our giant. This type of mass transfer in a binary is very common in the Universe, and it has a big impact on the life of our stars because it will change the three key characteristics mentioned above: the mass of star 1, the mass of star 2, and how far they are from each other. Yes! Moving mass from one star to the other will affect the orbit - it can make the stars come closer together or move apart, depending on some maths that you’ll be thankful I won’t discuss here.
But what if our giant star becomes so big that its outer layers reach the companion star… is that even possible? It sure is, and it is a crucial aspect of binary evolution. When a star becomes so large that its companion is effectively swimming in its outer layers, we call that the “common envelope” phase. And it is anything but a relaxing swim. As the unfortunate companion ploughs through the envelope of the giant, it feels a drag force that slows it down, and slowing down on an orbit means only one thing: you are going in. This common envelope phase tremendously shortens the distance between the two stars, and sometimes (often?) it even leads to a full merger. This fatal option can be avoided if the large envelope engulfing the second star is ejected sufficiently early (effectively being “blown” away by the companion star ploughing through it).
So the lives of stars can be very tumultuous. The work of stellar genealogy is figuring out which parent stars and which series of events (such as mass transfer or common envelope phases) had to happen for us to see a particular star, or indeed, a particular explosion.
Types of binary interactions that a star can undergo over the course of its life
(Stable) mass transfer:
When a star expands too much its outer layers start “pouring” down the gravitational well of the companion. Because the stars are spinning that material spirals in towards the other star (or object) and forms an accretion disc. This type of mass transfer can be quite stable over time - and if it is not it quickly turns into a common envelope.
Artist impression of an accretion system where a star transfers mass to a black hole Credit: ESA/Hubble. Copyright: CC BY 4.0
Contact Binaries:
Beta Lyrae is an example of a contact binary - where one star has expanded so much that the two stars touch. This amazing gif of the system was compiled with real telescope images and shows the stream of stellar material linking the two stars as they spin in their orbit. [Beta Lyrae - wikipedia]
Common Envelope:
Sometimes the star swells so much it will engulf its companion. It can be hard to visualise how much such a star can expand compared to the size of the system, so here you can see a gif of the evolution of the size and separation of two stars in one of the systems in our simulations. As you can see the Red Super Giant easily swallows its friend and as it does it draws it in, shortening their orbit. (Axes units: Solar radii)
As far as I am concerned, one of the most exciting discoveries of this century is that of the first observed Kilonova, charmingly named AT 2017gfo. You will probably have heard of supernova explosions, which in most cases are the result of the death of a massive star. This sounds like a pretty terminal outcome but it is anything ‘but’ - if you have a binary companion that is. When massive stars die in supernovae they typically leave behind what is called a Neutron Star - essentially its dead, collapsed, core. These are the densest stars in the Universe - in fact, if they were any denser they would turn into black holes and lose their shine. Now if our newly born neutron star has a binary companion, it too can hoover up its outer layers and plough through them in mass transfer or common envelope events. The life story of our binary does not end with a supernova. If the other star also ends its life creating a neutron star we now have a binary neutron star system. All we need for it to create a kilonova explosion is to wait for it to merge, and it can do so because the extremely dense stars create ripples in the fabric of space time as they orbit around each other [See this excellent video explaining gravitational waves with Spandex]. Making waves, gravitational waves even, requires energy and that energy is taken away from the orbit of the system: the two remnants slowly spiral in over the course of millions (billions!) of years.
By August 2017 astronomers had already detected gravitational waves from merging binary black holes, but the new signal detected on the 17th (called GW170817) was different: the masses were low, too low… at least for black holes. These were Neutron Stars, and they had merged. What followed was the most epic egg-hunt of our generation - a Kilonova explosion had happened, but where? Optical telescopes all over the world went looking and within a day of the merger, several teams had discovered a bright new flash of light in a galaxy called NGC 4993, about 130 Million light years away: the first unmistakable, undeniable, Kilonova had been uncovered.
Discovery of the First Neutron Star Merger
NGC 4993 (galaxy) and AT 2017gfo (kilonova - highlighted by the square). Credit: A. Levan/ESA
Kilonova explosions are rare, very distant, and they occur millions/billions of years after their parent star system was born. All we get as astronomers with a human, finite, life expectancy is a snapshot in time. In this case the snapshot is the death of the system. So how can we reconstruct the past of an object that exploded over 100 million light years away? We need two things: data and models.
Data was no issue - every telescope on Earth looked at this shiny new dot in the sky. Even better, it was so close to our Milky Way (everything is relative) that we have a beautiful view of its home galaxy - NGC 4993 - see image above. This is important because even if we can’t observe the system itself (on account of the fact that it exploded, for one) we can observe the light of the stars from the rest of the galaxy. And if you are an astronomer, light is gold. With the right model on hand, it can tell you a lot, such as the age and the chemical composition of the stars that make up this galaxy. Now stars are not born one at a time, but as groups and clusters, so if we can figure out the age and chemical make-up (so-called “metallicity”) of our galaxy, it tells us about the parent system that was born within it. In addition, we have information about our system when it died: the gravitational waves tell us about the masses of the neutron stars that merged.
So with all of this, we can figure out how long ago the parent system would have been born, what metallicity it was born with, and what final binary neutron star system it created. All we have to do is retrace every step of the billions of years that lie in between.
Now onto the models. Teams at the University of Auckland (NZ - led by Jan Eldridge) and University of Warwick (UK - led by Elizabeth Stanway) have been developing and using state-of-the-art simulations of how binary stars live and die in the Universe. The Binary Population And Spectral Synthesis (BPASS) models offer the broadest set of predictions in the field on what systems you expect to see in the cosmos, simulating their entire lives, telling you how many of each system you expect to see (their probability of occurring), and what they would look like from a distance. What is unique with BPASS is that you can use the models to match the light we get from NGC 4993 (to infer the age and metallicity) and you get the genealogy of each system simulated. This allows you to compare the final masses of the neutron stars that we “saw” merge to find the simulated stars systems that give you the right masses and collide at the right time. Once you find the matches, all you have to do is look inside your model to “rewind the tape” and reconstruct the stellar genealogy of your explosion.
Put this way it sounds pretty simple, but in practice…. it takes years to get it all running. To accomplish this full analysis I needed to create software pipelines to make the data, the models and the fitting codes work together. I also had to run additional statistical simulations to take into account how supernova explosions kicked the newly created neutron stars and changed the orbit of the binary systems (if you kick it too hard and you unbind the two stars you won’t get a merger!).
All-in-all, here are the coolest fact about the Genealogy of AT 2017gfo
The parents of the Kilonova were most likely born about 8 Billion years ago which is around what astronomers call “Cosmic Noon”: the period in the Universe when the most star formation took place (fun and scary fact, most of the stars in the Universe that will ever by formed have already been born - it’s all downhill from here).
The most likely parent system had a primary star between 11-25 the mass of the sun and a secondary star between 10-15 solar masses.
The other systems (about 15% of the probability) had a secondary star around 6 solar masses - that is too low to create a second neutron star! That means they had to gain a lot of mass from the first star in order to get big enough. They do this through stable mass transfer very early in the life of the first star - before it even becomes a Giant!
All our systems go through common envelope at least twice: once from the primary star, then from the secondary.
Some weirdo secondary stars went through common envelope TWICE! (So the system did it 3 times in total). But that is a very small fraction of systems (no more than 3%).
What is most exciting to me though is the software pipelines (“booooo boriiing!”). My hope is that what took me 2 to 3 years to do here will take only a few days next time we see a Kilonova. The gravitational wave detectors that allowed the discovery of GW170817/AT 2017gfo will turn back on in 2023 and run for 12 months. Then they will turn on again for even longer later in the 2020s, every time more powerful than the last. We will see more events like this one, and the type of analysis I did can be done systematically and quickly every time now. Also, it is worth mentioning that this stuff does not only apply to one type of explosion! If you see a supernova in a nearby galaxy and you have data on its home and neighbour stars, you are in business!
Here is to 3 years of hard work and many more explosive stellar deaths - CHEERS!