Light and gravitational waves don’t arrive simultaneously

There’s an important rule in relativity that — as far as we know — all objects must obey. If you have no rest mass as you travel through the vacuum of space, you absolutely are compelled to travel exactly at the speed of light. This is exactly true for all massless particles, like photons and gluons, approximately true for particles whose mass is tiny compared to their kinetic energy, like neutrinos, and should also be exactly true for gravitational waves. Even if gravity isn’t inherently quantum in nature, the speed of gravity should be exactly equal to the speed of light. At least, that’s a necessity if our current laws of physics are correct.

And yet, when we saw the first neutron star-neutron star merger in both gravitational waves and with light, the gravitational waves arrived first by a substantial, measurable margin: by almost 2 seconds. What’s the explanation? Even though the signal originated from 130 million light-years away, the distance shouldn’t matter; if the signals were generated at the same time, and they travel at the same speed and follow the same path, then they should’ve arrived at the same time, too.

Originally a puzzle with a few competing ideas surrounding it, we’ve now gathered enough data about these events and the rest of the Universe that one explanation stands out from the rest. Here’s the story of why we think that light’s arrival was “delayed” relative to the measured gravitational wave signal.

Illustration of a fast gamma-ray burst, long thought to occur from the merger of neutron stars. The gas-rich environment surrounding them could delay the arrival of the signal, but the mechanism that produces it could also cause a delay in the emission of the signal. Light and gravity should both travel, through the vacuum of space, at the same speed. We do not have information about how the neutron star-neutron star merger rate has evolved throughout cosmic history.

On August 17, 2017, the signal from an event that occurred 130 million light-years away finally arrived here on Earth. From somewhere within the distant, shell elliptical (or lenticular) galaxy NGC 4993, two neutron stars had been locked in a gravitational dance where they orbited one another at speeds that reached a significant fraction of the speed of light. As they orbited, they distorted the fabric of space owing to both their mass and their motion relative to the curved space through which they traveled.

Whenever masses accelerate through curved space, they emit tiny amounts of invisible radiation that’s invisible to all light-based telescopes: gravitational, rather than electromagnetic, radiation.

These gravitational waves behave as ripples in the fabric of spacetime, carrying energy away from the system and causing their mutual orbits to decay. As time went on, the two neutron stars began to inspiral, with gravitational waves carrying orbital energy away, causing these two objects to migrate closer and closer together. At a critical moment in time, these two stellar remnants spiraled so close to one another that they physically touched, and what followed was one of the most spectacular scientific discoveries of all-time.

This four-panel illustration of the inspiral and merger of two neutron stars shows how the amplitude and frequency of the gravitational waves both increase as the merger becomes imminent. At the critical moment of merger, the signal spikes, and then disappears behind the event horizon as a black hole is formed. Optical and other electromagnetic light may or may not be emitted as part of this process, but if a neutron star results, even briefly, from the aftermath of this merger, ejected material, including large quantities of heavy elements, should also be produced.

As soon as these two stars collided, the gravitational wave signal came to an abrupt end. Everything that the LIGO and Virgo detectors saw was from the inspiral phase up until that moment, followed by total gravitational wave silence. According to our best theoretical models, this was two neutron stars inspiraling and merging together, likely resulting in a remarkable end result: the formation of a black hole.

But then it happened. 1.7 seconds later, after the gravitational wave signal ceased, the first electromagnetic (light) signal arrived: gamma rays, which came in one enormous burst. From the combination of gravitational wave and electromagnetic data, we were able to pin down the location of this event better than any gravitational wave event ever: to the specific host galaxy in which it occurred, NGC 4993.

Over the coming weeks, light began to arrive in other wavelengths as well, as close to 100 professional observatories monitored the spectacular afterglow of this neutron star merger.

For the 2017 neutron star-neutron star merger, an electromagnetic counterpart was robustly seen immediately, and follow-up observations, such as this Hubble image, were able to see the afterglow and remnant of the event. For GW190425, the only other neutron star-neutron star merger seen in gravitational waves, no analogous electromagnetic signals have ever been observed.

On the one hand, this is remarkable. We had an event occur some 130 million light-years away: far enough away that light took 130 million years to travel from the galaxy where it occurred to our eyes. Back when the merger took place, planet Earth was a vastly different place. Feathered birds had been around for only 20 million years; placental mammals for 10 million. The first flowering plants were just beginning to emerge, and the largest dinosaurs were still yet to come into being: something that wouldn’t occur until 30 million years in Earth’s future.

For all that time, from then until the present, both the light and the gravitational waves from this event were journeying through the Universe, traveling at the only speed they could — the speed of light and the speed of gravity, respectively — until they arrived at Earth after a journey of 130 million years. First the gravitational waves from the inspiral phase arrived, moving the mirrors on our gravitational wave detectors by an incredibly small amount: less than a ten-thousandth of the size of an individual proton. And then, just 1.7 seconds after the gravitational wave signal ended, the first light from the event arrived as well.

An illustration of a very high energy process in the Universe: a gamma-ray burst. These bursts can arise when two neutron stars merge, and one was detected just 1.7 seconds after the gravitational wave signal from GW170817 ceased.

Immediately, this gave us the most impressive physical measurement of the speed of gravity ever: it was equal to the speed of light to better than 1 part in a quadrillion (10¹⁵), as it takes around four quadrillion seconds to make up 130 million years, and they arrived less than two seconds apart from one another. Prior to that, we had excellent theoretical reasons for knowing that the speed of gravity ought to equal the speed of light, but only had indirect constraints that the two were equivalent to within 0.2% or so. The improvement on our constraints of more than 12 orders of magnitude, with one single observation, represents the most significant leap from a single measurement of all-time.

Does this mean that the speed of gravity and the speed of light aren’t quite equal, then? That perhaps either gravity moves slightly faster than c, the speed of light in a vacuum, or that light itself might actually move a tiny bit slower than c, as though it had a tiny but non-zero rest mass to it?

That would be an extraordinary revelation, but one that’s highly unlikely. If that were true, light of different energies (and wavelengths) would travel at different speeds, and the level at which that would need to be true is much too large to be consistent with observations.

The longer a photon’s wavelength is, the lower in energy it is. But all photons, regardless of wavelength/energy, move at the same speed: the speed of light. This is, surprisingly, irrespective of the motion of the observer relative to light; the speed of light is the same for all observers.

In simpler terms, if light had a non-zero rest mass, and that mass were heavy enough to explain why gravitational waves arrived 1.7 seconds earlier than light after traveling 130 million light-years across the Universe, then we’d observe radio waves traveling significantly slower than the speed of light: too slow to be consistent with what we’ve already observed.

But that’s okay. In physics, we don’t have any problem considering all possible explanations for an observed puzzle. If we’re doing our jobs correctly, then we’ll be able to consider every fathomable explanation, hold the data up against each one of them, and all of them will be ruled out except for one. The challenge is to find the correct explanation that fits all facets of our observations, and is still powerfully predictive for what signals should appear versus the ones that actually showed up.

And we think we have!

The key is to think about the objects that are merging together, the physics at play, and what signals they’re likely to produce. We’ve already done this for the gravitational waves, detailing how they’re produced during the inspiral phase and cease once the merger takes place. Now, it’s time to go a little deeper and think about the light.

A gamma-ray burst, like the one depicted here in an artist’s rendition, is thought to originate from a cataclysmic event occurring inside a dense region of a host galaxy, possibly surrounded by a large shell, sphere, or halo of material. That material will have a speed of light inherent to that medium, and individual particles that travel through it, although always slower than the speed of light in vacuum, might be faster than the speed of light in that medium. In the instance of a kilonova, light slows down passing through the material surrounding a pair of merging neutron stars.

Up until these two neutron stars touched, there was no “extra” light produced. They simply shone as neutron stars do: faintly, at high temperatures but with tiny surface areas, and completely undetectable with our current technology from 130 million light-years away. Neutron stars aren’t like black holes; they aren’t point-like. Instead, they’re compact objects — typically somewhere between 20 and 40 kilometers across — but denser than an atomic nucleus. They’re called neutron stars because they’re about 90% neutrons by composition, with other atomic nuclei and a few electrons at the outer edge.

When two neutron stars collide, there are three possibilities that can result. They are:

1. you can form another neutron star, which you’ll do if your total mass is less than 2.5 times the mass of the Sun,
2. you can form a new neutron star briefly, which then collapses into a black hole in under a second, if your total mass is between 2.5 and 2.8 solar masses (dependent on the neutron star’s spin),
3. or you can form a black hole directly, with no intermediate neutron star, if your total mass is greater than 2.8 solar masses.

We knew that when two neutron stars merge, as simulated here, they can create gamma-ray burst jets, as well as other electromagnetic phenomena. But perhaps, above a certain mass threshold, a black hole is formed where the two stars collide in the second panel, and then all the additional matter-and-energy gets captured, with no escaping signal. Determining the mass boundary between where neutron stars and black holes can form is one of the goals of modern gravitational wave astronomy.

From the gravitational wave signal that arose from this event, officially known as GW170817, we know that this event falls into the second category: the merger and post-merger signal existed for a few hundred milliseconds before disappearing entirely all in an instant, which indicates that a neutron star formed for a brief time before an event horizon formed and engulfed the entire thing.

But nevertheless, light still got out. The next question was, simply, how?

How was the light that we observed generated? Again, there were three possibilities that we could think of.

1. Immediately, as soon as the neutron stars touch, by processes that occur on their surfaces.
2. Only after material gets ejected, where it collides with any surrounding material and produces light from that.
3. Or from the interior of neutron stars, where reactions generate energy that only gets emitted once it propagates to the exterior.

In each scenario, gravitational waves travel unperturbed once the signal is generated, but light takes an extra amount of time to get out.

In the final moments of merging, two neutron stars don’t merely emit gravitational waves, but a catastrophic explosion that echoes across the electromagnetic spectrum. Whether it forms a stable neutron star or a black hole (like the 2019 merger), or a neutron star that then turns into a black hole (like the 2017 merger), will depend on factors like the total mass of the predecessor neutron stars and their combined spin.

If it’s the first option, and neutron star mergers generate light as soon as they touch, the light gets emitted immediately and therefore must be delayed by passing through the environment surrounding the neutron star. That environment must be rich in matter, as each fast-moving neutron star, with charged particles on their surfaces and intense magnetic fields, is bound to strip and eject material from the other one.

If it’s the second or third option, merging neutron stars generate light from their mergers, but that light only gets emitted after a certain amount of time has passed: either for ejected material to smash into the circumstellar material or for the light generated in the neutron star interiors to reach the surface. It’s also possible, in either of these cases, that both “delayed emission” and “slowed arrival by surrounding material” are at play.

Any of these scenarios could easily explain the 1.7s delay of light’s arrival with respect to gravitational waves. But on April 25, 2019, we saw another neutron star-neutron star merger in gravitational waves, which was more massive than GW170817. No light was emitted of any type, disfavoring the first scenario. It looks like neutron stars don’t generate light as soon as they touch. Instead, the emission of light comes after the emission of gravitational waves.

Artist’s illustration of two merging neutron stars. The rippling spacetime grid represents gravitational waves emitted from the collision, while the narrow beams are the jets of gamma rays that shoot out just seconds after the gravitational waves (detected as a gamma-ray burst by astronomers). Mass, in an event like this, gets converted into two types of radiation: electromagnetic and gravitational. About 5% of the total mass gets expelled in the form of heavy elements.

With only two direct detections of merging neutron stars through the emission of gravitational waves, it’s a testament to how incredibly precise the science of gravitational wave astronomy has become that we can reconstruct all we have. When you add in the electromagnetic follow-up observations from the 2017 event that also produced light, we’ve definitively shown that a large fraction of the heaviest elements in our Universe — including gold, platinum, iodine and uranium — arise from these neutron star mergers.

But not, perhaps, from all neutron star mergers; perhaps it’s only the ones that don’t immediately form a black hole.

Either ejected material or reactions in the neutron star’s interior is required to produce these elements, and hence, the light associated with a kilonova explosion. That light is only produced after the gravitational wave signal has ended, and may further be delayed by having to pass through the circumstellar material. This is why, even though light and gravity both travel exactly at the speed of light in a vacuum, the light we saw didn’t arrive until nearly 2 seconds after the gravitational wave signal ceased. As we collect and observe more of these events, we’ll be able to confirm and refine this picture once and for all!