This past month, I have been working hard with my colleagues to understand an exciting new discovery – the first (definitive!) kilonova.

Background Information

Kilonovae occur when either two neutron stars or a neutron star and black hole collide. They are thought to be the birth place for many of the heavy elements in our universe, known as the r-process. You can read more about the science behind these events on our website,, which I created with Peter Williams and other members of our team.

On August 17, 2017, Advanced LIGO and Virgo detected a small ripple in space time originating from the merger of two neutron stars. Our team, along with many others around the globe, began following the event in a multi-wavelength campaign spanning from X-rays to radio.

A Multi-wavelength View of GW170817

We have presented eight initial papers on this new kilonova, which can be found here or in ApJL.

The broad wavelength view allowed us to follow several physical components of the merger itself. We tracked radioactive r-process decay with optical and near infrared light curves, and found more specific signatures of r-process elements using spectroscopic datasets. The X-rays and radio tracked the afterglow of an off-axis short GRB at late times. We compared the GRB itself to many others, finding that this even was mysteriously dimmer by order of magnitude. Using HST images of the host, we were even able to constrain the age of the neutron star binary and track the history of its host.

Spectra video

A One-Zone, Semi-Analytical Model

I specifically worked on developing and fitting a semi-analytical model for our paper lead by my bud Phil Cowperthwaite. Earlier this year, I published a paper which broadly explored optical light curves of many classes of transients using the new open-source code MOSFiT. During that time, I implemented a variation of the toy kilonova model presented by Brian Metzger last year. This ended up being very lucky for our team, because we were able to use the powerhouse MOSFiT to run an MCMC fitting routine on our awesome optical and NIR photometric dataset reduced largely by Phil. We were also able to compare our light curves to other more simple models (like the radioactive decay of Nickel) to rule out less exotic explanations for the data.

kilonova light curve and model

In short, we found that the ejecta mass of the event was overall larger than expected adn that the kilonova had to be explained by multiple components of varying Lanthanide-fraction. The large mass implies a highly asymmetric merger – a prediction eventually confirmed by the LIGO data!

I also made a order-of-magnitude calculation in the paper which compared the mass of our event to the overall r-process abundance in the Milky Way. We found that our event over predicts the yields in our galaxy, but not by too much.

There is still a lot left to be done with this object and the new field as a whole!