NICER captures the likely birth of a compact object in an unusual cosmic explosion

The discovery of a rapid 4.4 millisecond X-ray quasi-periodicity from the extragalactic explosion AT2018cow combined with previous multi-wavelength studies suggests we may have identified a newborn compact object (a black hole or neutron star) immediately after a supernova.

Like Comment
Read the paper

It was the fall of 2018 when I first heard of the optical explosion named AT2018cow. Sjoert van Velzen, a close collaborator and friend, told me about this extraordinary optical event that had happened in a nearby dwarf spiral galaxy situated at a distance of just 60 megaparsecs. The properties of this explosion dubbed “the cow” were fascinating. For starters, it rose to its peak brightness in a matter of 2 days, when it became one of the brightest known supernovae. In comparison, typical core-collapse supernovae take weeks and are much fainter. Intriguingly, its early optical spectra did not match any of the known classes of supernovae. On top of all these extreme qualities it also had bright X-rays variable on timescales of tens of hours–which is again unusual for normal supernovae.

If AT2018cow originated from the core collapse of a massive star then the fact that it is roughly 100 times brighter than a typical core-collapse stellar explosion suggests that there is an additional source of energy, perhaps a central compact object or shocks produced via interaction of the explosion with dense ambient medium. If not a stellar explosion then the alternative hypothesis is that it could be a member of a long-sought class of intermediate-mass black holes weighing 10,000-100,000 solar masses and tidally shredding a white dwarf star. My interest in the cow peaked after that conversation with Sjoert. I remember thinking X-ray variability should be able to differentiate between all these major scenarios, and that data from the Neutron Star Interior Composition Explorer (NICER), an astrophysics payload on the International Space Station, if available, would be ideal for this purpose. 

Luckily, Jon Miller of the University of Michigan Ann Arbor had already requested Target of Opportunity time to observe this object with NICER. Both Keith Gendreau and Zaven Arzoumanian (NICER’s Principal Investigator and Science Lead, respectively) were also intrigued by this peculiar cosmic explosion, so they monitored it with NICER over a two month period following its discovery. This meant that AT2018cow’s NICER data were already available for anyone to take a deep dive into.

So as soon as I got back home, I downloaded the entire NICER data set and extracted an average power density spectrum, which is a tool to calculate how the source is varying on a range of timescales. If there are any regular brightness variations on a certain timescale, this would manifest as excess power at the associated frequency in the power density spectrum. To my surprise, there was excess power near 224 Hz or 4.4 ms (see Figure below).

Average power density spectrum of AT2018cow showing the signal near 224 Hz. The red curve is the best-fit model (constant+Lorentzian).

As with many scientific results, the discovery of the signal itself was the easy part. By far, the majority of the work was to 1) estimate the chance of this being a statistical fluke, which requires an accurate characterization of noise, and 2) ensure that this signal was not an instrument anomaly. After several months of extensive analysis, conjectures, and discussions with folks who built the instrument, we concluded that the probability of seeing a signal as strong as the one we found in the noise is about 0.02%, i.e., the odds of this being a statistical fluke is 1 in 5000 (or 3.7σ). We also performed numerous tests to rule out various non-astrophysical origins for this signal. 

A 4.4 millisecond modulation of the X-rays sets a strict upper limit on the physical size of the region producing this signal. Intuitively, imagine a clump of hot matter regularly brightening and getting dimmer on average every 4.4 milliseconds. For this clump to be able to vary as a whole, information should be able to propagate across the length of the clump in much less than 4.4 ms. Otherwise, it would not be able to produce the regular modulation. This means that the size of the region must be less than the speed of light, i.e., the maximum speed at which information can travel, multiplied by 4.4 milliseconds. Thus, the signal we detected sets a strict upper limit of about 100 million centimeters on the size of the object producing X-rays in AT2018cow. This compact size argues for the presence of a black hole or a neutron star and argues against a shock origin for X-rays.

Now, we can turn this size estimate into a mass limit because we know, from general relativity, the maximum mass one can pack within a certain region before it collapses into a black hole. Assuming a maximally spinning black hole the upper limit on the mass that can be contained within 100 million centimeters is 850 solar masses. If the black hole is not spinning, this upper limit reduces by a factor of 6. It is important to realize that this mass upper limit is based on several extreme values: 1) that information travels at the speed of light, 2) it would take as long as 4.4 milliseconds, and 3) the central compact object is spinning at the maximum value allowed by nature. In reality, it is unlikely that so many extremes are at play and as a result the upper limit is probably several factors lower than 850 solar masses. In any case, this mass limit rules out intermediate-mass black holes. An illustration of our result can be seen below and you can access the full paper by clicking the "Read the Paper" button at the top of this blogpost. 

It still remains a puzzle whether the compact object at the heart of AT2018cow is a black hole or a neutron star. The most straightforward way to distinguish between the two would be to identify a pulsar spin—the strictly periodic flash of a spinning “lighthouse” beam from a neutron star. It is possible that the pulsar spin period is approximately  4.4 milliseconds and is perhaps evolving over the two-month NICER monitoring period, resulting in a quasi-periodic signal. To identify the pulsar period and its evolution, one would need to perform a search over several possible models of spin evolution and their respective parameters. This would require a significant amount of computation power, which one can, in principle, carry out if Graphics Processing Units (GPUs) are accessible. We intend to perform such a search in the future. 

Whichever scenario for the central compact engine is correct, the presence of a rapid modulation of X-rays is fascinating and invites speculation. In the context of a black hole, a 224 Hz signal could result if some debris from the exploded progenitor star fell back into the black hole, temporarily forming an accretion disk that, perhaps, wobbled with this characteristic frequency. Or in the neutron-star picture, we may have seen the effects of cooling and shrinking of its ultra-dense interior, following the fiery creation of this stellar cinder. Either way, it seems that we have witnessed, for the first time, the birth throes of an exotic and extreme object. 

 While doing the analysis, performing the checks, interpretation and writing the paper was a huge effort by all of us, I am very excited for the future. AT2018cow may be just the beginning of what is to come, as many more objects like it are expected in the era of deeper all-sky surveys. 

Dheeraj Pasham (DJ)

Research scientist, MIT

My area of expertise is high-energy astrophysics/time domain astronomy. I am excited about pretty much all kinds of cosmic explosions especially those involving black holes (small, medium and large) and neutron stars.