A binary star and a big explosion: SN 2017gci
Can the BPASS code natively predict the progenitors of (some) super luminous supernovae? Yes it can.
Super bright Supernovae
I have been obsessed with super luminous supernovae (SLSNe) since the beginning of my PhD and, although I am not in the field, I have been keeping an eye on these explosive ladies. In case you’re not aware, SLSNe are special: they are brighter and more powerful than typical supernovae — so much so that the typical mechanism that powers the bulk of the supernvoae which arise as a result of the death of massive stars just cannot explain them. You simply can’t squeeze enough energy out of the core collapse (e.g. see Figure 1 Inserra 2019).
So how do you power a superluminous supernova then?
There are a few options, here are some of the most commonly invoked mechanisms:
Pair-instability supernova
Magnetar spin-down injects extra energy into the ejecta
Circumstellar Medium in the way of the ejecta gives an extra boost when the two interact
If you’re keen to learn more there is a recent review by Gal-Yam (2019) you can check out — but what you need to know here is that the explosion and powering mechanism is still debated. The magnetar models have been rather successful so far, but realistically different explosions might have different explanations, and sometimes even require a combination!
Figure from Fiore et al. 2021
That is the case for SN 2017gci (Fiore et al. 2021), an event that peeked my interest when I saw the pre-print hit the arxiv at the end of 2020. The title is really what did it for me “[…] a nearby Type I Superluminous Supernova with a bumpy tail”. I had not yet heard of bumpy tails in SLSNe and so I just had to take a look: this is what the lightcurve looked like (see figure below): Notice how the data around 100 days don’t just decay smoothly?
The most likely case is that in addition to having a magnetar powered explosion (as shown by the yellow curve) we have some circumstellar material (CSM) around the progenitor. If that CSM is in shells or has uneven density throughout, you’ll end up with extra light of the bumpy variety.
The question, once again, is how?
In particular, assuming that this CSM comes from the progenitor, its expulsion from the envelope of the dying star must have been particularly well timed — see, this material contains hydrogen, but the supernova explosion did not. Given the speed of the ejecta (~8000 km/s), the time at which we see the bumps and the typical wind speed for massive stars (~1000 km/s), you can deduce that the progenitor of SN 2017gci still contained hydrogen mere decades before the supernova, but had managed to lose it all by the time it died.
When I first saw this, my initial reaction was “BINARIES!”. Binary interactions are very good at stripping stars, and I figured that if our SLSN progenitor was undergoing mass transfer (being eaten up by its companion, see the cartoon below) or a common envelope phase (where the outer parts of the two stars have merged but the cores remain separate), then the material that can be lost in these events could give us the necessary ingredients to lose the hydrogen layers just in time to explain SN 2017gci.
And so I went looking for this….
A mini universe in your computer
So how do you find a star that has all the right characteristics? You comb through your simulations, or in my case my boss’s simulations. I used my code hoki to compile the whole library of stellar models in BPASS into tables that are easy to search through in python [download here], and then I used the information in the Fiore et al. (2021) and in the literature to create a set of criteria that would need to be matched.
BPASS is a state of the art stellar evolution code: a mini universe in a box. It contains a million solar masses split across millions of stars. These stars sometimes live alone, sometimes have partners (binary systems), in the proportions we see in the cosmos ( Eldridge et al. 2017) . They are followed throughout their lives and the code records over a hundred properties for each star such as the brightness, temperature, mass, radius, etc…
All I need to do is look at what each star looks like when it dies to see if it matches the criteria in the table below. Once I find a match I can rewind the tape and see what happened over the course of its life!
| Property | Matching Criteria | Number of models | |||
|---|---|---|---|---|---|
| Metallicity | 0.006 | 21.598 | |||
| Mass (ZAMS) | >10 Msun | 9476 | |||
| Mass remnant | < 2.43 Msun | 9233 | |||
| Mass ejecta | 8-13 Msun | 1863 | |||
| Hydrogen mass fraction | < 0.01 | ||||
| Hydrogen Mass | < 0.001 Msun | 761 | |||
| t_csm | < 150 years | 2 |
Table 1: Set of criteria applied to the BPASS models. Explosion energy used here is 1e52 ergs.
t_csm is the time at which the hydrogen content drops below the limits indicated in the two rows above.
It’s a match!
There were two models in the BPASS/hoki tables that ticked all the boxes — both with a 30 solar mass zero-age maine-sequence star. The only difference is in the companion: one has a 12 solar mass secondary star, the other is the secondary to a 25 solar mass black hole primary. The former is over 110 times more likely to occur in nature and so that is the one I focus on, plus, from the point of view of the progenitor of SN 2017gci, they are virtually identical.
The life of this star is summarised by the figure below and as you can see it’s not really what I expected: The binary interactions don’t happen just before the supernova at all!
Now don’t get me wrong, these phases of common envelope and mass transfer are crucial to our star losing its hydrogen at the right time, even though they happen roughly 400 000 years before the boom — see, BPASS also comes with single-star-only simulations so that you can compare, and the single stars cannot match all the criteria I describe in Table 1. Either they lose their hydrogen way too early or they don’t lose it at all.
But the binary interactions on their own do not write the whole story of that suggested CSM around SN 2017gci — we need something else, some form of mass loss at the 11th hour. And this happens in the last few thousand years as the stellar winds strengthen after helium runs out in the core of our dying star. The luminosity output skyrockets and the mass loss rate reaches a whooping 2e-5 Msun/year - that is over 1.5 earth masses lost to space EVERY YEAR… until 50 years before the supernova, when it finally runs out of hydrogen and its radius plumets, becoming smaller than our Sun but weighing 12 times as much.
What a beast!
The coolest things about this
1) This progenitor appears naturally in BPASS from our current understanding of stellar evolution!
2) This is a “low mass” pathway that covers a mass regime not encompassed by previous scenarios of other SLSNe with bumpy tails! It helps explain new areas of parameter space.