Is Einstein Still Right?

by Clifford M. Will

  The “discovery” event GW150914 was not the only one recorded during that first observing run, called O1 by the LIGO collaboration, which ran from mid September 2015 until 19 January 2016. The second event came on 12 October, but it was so weak that the statistical probability that it was a fluke was just once every three years, much larger than the probability of less than once every 200,000 years for the discovery event. Because of this, the LIGO collaboration decided not to announce this event as a proper detection, but rather as a “candidate” event which could not be confidently confirmed as an actual gravitational wave. A later reanalysis would promote this to a reasonably confident detection of a 23 and 14 solar mass black hole merger. The third signal was detected in the early hours of 26 December 2015, Greenwich time, although it was still Christmas day in the US. It was called the “Boxing Day event,” after a tradition in England and Commonwealth countries of giving tradesmen, postal workers or staff a box containing cash or a small gift on the day after Christmas. This event was quite similar to the first one in that it was produced in the merger of black holes, located roughly the same distance from Earth, but with smaller masses (14 and 7 solar masses). Consequently, this event was not as loud as the first one, but still loud enough to be statistically significant.

  The events we just described are the only ones LIGO detected during its first observing run. Following some improvements in the sensitivity of the detectors, the second observing session, O2, ran from November 2016 to the end of August 2017, and the haul of detections was impressive. A high-mass binary black hole inspiral (31 and 20 solar masses) was detected on 4 January 2017, and a moderate-mass inspiral (11 and 8 solar masses) was found on 8 June. Four more black hole inspirals were detected by LIGO during O2 between late July and the end of August, including an event with two huge black holes, weighing in at 50 and 34 solar masses, producing a final monster hole of 80 solar masses. This source was also the most distant by far, at nine billion light years. The other three detections were quite similar to the discovery event GW150914 in mass, distance and other characteristics.

  Ironically, these four detections made during the summer of 2017 were not announced until December 2018, because of what occurred during the fateful week of 13 August 2017. Two detections made three days apart during that week were so exciting that everything else was pushed aside in order to focus on these finds.

  As we discussed in Chapter 7, the Virgo gravitational wave detector in Italy was about a year behind LIGO in the various stages of construction, commissioning and upgrading. But on 1 August it joined LIGO in a three-way session, scheduled to last for three weeks, until the end of O2. And on Monday 14 August 2017, the three instruments heard the gravitational waves emitted by another binary black hole merger, with masses in the 25 to 30 solar mass range. But the triple detection allowed for something new: pinpointing the location of the source in the sky.

  How can these detectors determine where the waves came from? Just as in navigation using GPS, discussed in Chapter 2, the answer is triangulation, exploiting the arrival time of a signal. In GPS, it is the arrival time at the user of signals from multiple GPS satellites that allows the user to determine her location. In gravitational wave triangulation, it is the arrival of the signal from a single source at different detectors. For two LIGO detectors, we saw in Figure 8.3 how measuring the difference in arrival times at the two sites allows you to determine that the source lies somewhere on a circle on the sky (except for the special case where the source lies precisely along the line joining the two detectors). But if there is a third detector, such as Virgo, then in exactly the same manner, the difference in arrival time between Virgo and, say LIGO-Livingston, gives a second circle on the sky (the time difference between Virgo and LIGO-Hanford gives redundant information).

  There are three possibilities for these two circles. The first is that they do not intersect at all, in which case the event would have to be rejected as a candidate gravitational wave. This is analogous to the case where the time difference between the arrival at two sites is larger than the light travel time between the sites. The second possibility is that they intersect in two places, as illustrated in Figure 8.7. In this case, since the source must reside somewhere on both circles, it must therefore be at one of the two intersection points, either A or B. The third is a very special and lucky case in which the circles just touch each other at a single point, giving a single location for the source on the sky. In reality, these circles are really bands of some width, reflecting the uncertainties that are inevitable in noisy and imperfect data. And exploiting additional details of the response of the detectors to an impinging wave, it is possible to exclude parts of each circle as being less likely to correspond to the source’s location. Thus, while data on the prior LIGO-only detections could confine the sources at best to large elongated banana-shaped regions on the sky, the data on the LIGO–Virgo source GW170814 confined it to an oval region in the southern sky about the size of a major league baseball held at arm’s length.

  Figure 8.7 The arrival-time delay between two detectors fixes the source to lie on a circle in the sky. The delay between one of those and a third detector fixes the source on another circle. If the circles intersect, the source must lie at one or other intersection point (A or B).

  Three days later, on 17 August, another signal passed the Earth, reaching Virgo first, then LIGO-Livingston 22 milliseconds later, and LIGO-Hanford 3 milliseconds after that. This signal was quite different from all the black hole inspiral signals detected to that point. Instead of a very short, rapidly changing chirp–merger–ringdown signal, as shown in Figure 8.2, the signal was detected for a whopping 100 seconds, and appeared pretty boring, something like the signal shown schematically in Figure 8.8. In fact, Figure 8.8 illustrates only about the final quarter of a second of the signal (to picture the whole signal you have to imagine it continuing about 200 page widths to the left). The signal was a regular undulating wave, suggesting that the source is a binary system. Its frequency was roughly the same as for the black hole mergers, around 100 hertz, suggesting that the bodies are revolving around each other very fast. But compared to the black hole signals, where the changes in size and frequency of the wave could be seen over a few orbits, here the orbit-by-orbit changes are minuscule, suggesting that the rate of leakage of energy into gravitational waves is tiny. This indicates that the masses of the two bodies are much smaller than the masses in the black hole inspirals. All of this pointed to an inspiral of two neutron stars. The event was denoted GW170817.

  Figure 8.8 Gravitational wave of a neutron star binary inspiral. Only the last quarter of a second of the wave is displayed. The observed wave GW170817 lasted 100 seconds.

  In addition, while the signal was detected in both LIGO instruments, it was barely detected by Virgo. While it was possible that the Virgo detector was somehow malfunctioning at that particular moment (which would have been strange, since it worked perfectly three days earlier), the more likely explanation was that the source was at a location in the sky close to one of the four directions for which Virgo is “deaf” (see Figure 8.5). Because of that unlucky alignment, the Virgo interferometer did not respond as strongly to the signal as it might otherwise have. This useful information, combined with the circle in the sky inferred from the arrival times at the two LIGO detectors, gave a decent sky location for this source.

  Meanwhile, orbiting 534 kilometers above the Earth, detectors on board the Fermi Gamma-ray Space Telescope detected a burst of gamma rays in its routine sweep of the sky. The gamma rays arrived 1.74 seconds after the end of the gravitational wave signal. Fourteen seconds later, even before the automated software at LIGO and Virgo had fully registered what they had detected, Fermi issued an automated alert to astronomers worldwide (including the LIGO–Virgo team) so that follow-up observations could begin. This detection was denoted GRB170817 (for gamma ray burst). Forty minutes later, LIGO–Virgo issued its own worldwide alert, noting the near coincidence in time bet
ween the gravitational wave signal and the gamma ray burst. Five hours after the gravitational wave event, they had localized the source to be in a region of the sky about 30 square degrees in angular size, and at a distance of 130 million light years, good to about 20 percent. These observations meant that the source was somewhere within a three-dimensional cube in space containing about 49 galaxies. The Fermi observation was consistent with this, but was not accurate enough to pinpoint the specific galaxy from which the gamma rays had originated. About 12 hours after the initial detections, a team using the Swope Telescope at Las Campanas Observatory in Chile detected a counterpart signal in the visible band, and identified the host galaxy as NGC 4993.

  The galaxy NGC 4993 is not particularly interesting. Discovered in 1789 by astronomer Wilhelm Herschel, it is an elliptical galaxy in the constellation of Hydra. Unlike the Milky Way or the Andromeda galaxies, which are rotating galaxies with spiral arms and a dense core, an elliptical galaxy is more egg shaped, with stars that orbit almost randomly around its center. NGC 4993 is about the size of the Milky Way, and like most galaxies it hosts a supermassive black hole at its center, which in this case has a mass of roughly 80 to 100 million solar masses. It also shows evidence that it merged with another galaxy about 400 million years ago. So, apart from hosting humanity’s first ever detected neutron star inspiral and merger, NGC 4993 is a fairly run-of-the-mill galaxy.

  Ironically, while all this was happening that August morning, Nico was hosting a workshop at the eXtreme Gravity Institute at Montana State University in Bozeman, called “eXtreme Gravity meets eXtreme Matter.” The main purpose of this workshop was to bring experts together to discuss the science one would be able to extract once LIGO–Virgo detected gravitational waves from the merger of two neutron stars. This made it extremely difficult for half the attendees (who belonged to the LIGO–Virgo collaboration and had seen both alerts) to participate in the workshop discussions, as they had to abide by LIGO secrecy rules. How they managed to hold their excitement at bay and not spill the beans is a mystery. Those constraints would be lifted only months later, after the event was confirmed as a true detection.

  What followed the LIGO–Virgo and Fermi alerts was one of the most extraordinary observational campaigns in the history of astronomy. Over the next thirty days, follow-up observations were made in every band of the electromagnetic spectrum, using telescopes on the ground and in space. Thirteen separate teams made gamma ray and X-ray observations. The Hubble Space Telescope made observations in the ultraviolet, visible and infrared. Thirty-eight teams observed in the visible band, while twelve worked in the infrared band. Fifteen teams made observations in the radio band. Three teams even looked for signals of neutrinos (not surprisingly, given the distance of the source, they saw none). On 20 October 2017, Astrophysical Journal Letters published a summary of what all the observations had yielded during those first two months in what was being dubbed “multi-messenger” astronomy. The paper had 3,500 authors, of whom 1,100 were in the LIGO–Virgo collaboration and the rest were associated with astronomical projects. Virtually every astronomy department or center in the world was represented in the 953 listed institutes. A year after the detection of GW170817 and GRB170817, observations of the electromagnetic radiation from the source continued in many wavelength bands, and almost a hundred papers had been published in a range of astronomy journals, along with around fifty in physics journals.

  We might be forgiven for pointing out a certain irony in all this. In the early 1990s, when the US government was trying to decide whether to give the go-ahead for major funding to begin the construction of LIGO, many prominent American astronomers lobbied vigorously against it. The astronomers’ arguments roughly boiled down to some combination of: it won’t work; it will only detect gravitational waves, which we already know exist (see Chapter 5); it’s just a physics experiment with nothing to do with astronomy; and it is too expensive. One astronomer, J. Anthony (Tony) Tyson, reported that an informal poll he conducted in early 1991 of seventy astronomers ran four to one against LIGO. In March 1991, Cliff testified in favor of LIGO before a US House of Representatives Science subcommittee, alongside Tyson who, while supportive of LIGO in general, testified against construction funding at that time. What a difference a detection makes!

  What did we learn from this multi-messenger source? Detailed analysis of the gravitational wave signal indicated that the masses, around 1.5 and 1.3 solar masses, were quite consistent with those of known neutron stars. The gamma ray burst showed that it could not have been two black holes, because you need hot matter to get such high-energy radiation, and black holes are pure spacetime. A “mixed merger” of a neutron star and a black hole could not be definitively excluded, although it was hard to imagine where such a low-mass black hole would come from.

  The association of a neutron star merger with a burst of gamma rays resolved a long-standing mystery. Gamma ray bursts have been observed and studied since the 1960s, when the US Vela satellites accidentally detected the first bursts. These satellites had been deployed in the middle of the Cold War by the US military to investigate whether the Soviet Union was testing nuclear weapons in space. Gamma rays are a byproduct of such nuclear explosions, and indeed the Vela satellites detected many bursts of gamma rays. But the bursts did not have the characteristics of those emitted by nuclear bombs, and appeared to be coming from far outside the solar system.

  Within a few years, the study of these mysterious gamma ray bursts exploded. In 1991, the Compton Gamma Ray Observatory (CGRO) was launched, and over the next nine years it observed and localized around 2,700 gamma ray bursts (almost one per day), finding that they were not coming from any preferred direction, and thus suggesting an extragalactic origin. Moreover, astronomers deduced that the bursts came in two rough classes based on their duration: “short” bursts had an average duration of about 0.3 seconds, while “long” bursts lasted 30 seconds on average. Although the short bursts comprise only about 30 percent of all observed gamma ray bursts, they were particularly interesting. Astronomers realized that there was no correlation between these short bursts and supernovae, ruling out the latter as possible progenitors. They also realized that most of the short bursts came from elliptical galaxies where there is an underabundance of massive stars, which are needed for supernovae.

  Theoretical arguments made the short bursts even more intriguing. Imagine that the source that produces these short bursts of gamma rays is a ball or blob of matter of some size. Imagine that the burst is caused by an explosion of the blob, producing a flash of light that is directed toward us. The duration of the burst must be related to the finite extent of the source: we first see the light emitted from the region of the source closest to us, and later we see radiation emitted from other regions farther away from us. Since the entire duration of the burst is about 0.3 seconds, and since gamma rays travel at the speed of light, we conclude that the emitting region must be of the order of the time duration multiplied by the speed of light, which comes out to roughly 100,000 kilometers, or eight times Earth’s diameter. And because of the tremendous energy produced in these bursts, there had to be an enormous amount of matter contained in such a small region of space. Whatever produced these short bursts had to involve dense compact objects such as neutron stars, as Russian physicist Sergei Blinnikov and collaborators had theoretically predicted back in 1984.

  By 2005, astrophysicists began to suggest that the short bursts were the result of either neutron stars merging with each other or of a neutron star merging with a small black hole. But there was no way to prove this, since no light would be detected from the merging pair before the short gamma ray burst started. The LIGO–Virgo observations provided the missing piece of the puzzle, unequivocally proving that at least one of the possible progenitors of short gamma ray bursts is the merger of neutron stars. This observation also validated other ingredients of the model, such as the fact that the bursts are emitted in a very narrow cone, which we detect on Earth
only if the cone happens to be pointing in the right direction. This, in turn, suggests that many more short gamma ray bursts must be occurring in the universe, but with their emissions beamed in cones that do not point toward Earth. Gravitational wave observations will be able to determine precisely how many such events occur in the universe, since gravitational waves are not emitted narrowly in a cone, but rather more or less equally in all directions.

  These short gamma ray burst models also claimed to answer a question asked by anybody who has ever bought a wedding ring or examined the insides of a cell phone: where does gold come from? Today we have a very good idea of the origin of the key elements in nature. About three minutes after the big bang, around 20 percent of the primordial hydrogen was converted via nuclear fusion to helium, along with a sprinkling of lithium, a process known as big bang nucleosynthesis. Stars continue the process, converting more of their hydrogen into helium, but also extending the fusion process to elements such as carbon, nitrogen and oxygen, so crucial for the life forms that inhabit Earth. Very massive stars also produce those elements, but when they explode as supernovae, they produce elements up to the so-called “iron group” (iron, manganese, cobalt, nickel) and spew them into interstellar space, later to be incorporated into planets such as Earth. Unfortunately it is not so easy to produce elements heavier than iron. Many elements above iron in the periodic table are unstable, decaying to other elements by various radioactive processes, and they all have many more neutrons than protons in their nuclei. Nuclear physicists had developed a set of chain reactions, known as “r-process nucleosynthesis” (“r” meaning rapid, not the most imaginative of names), that could in principle produce heavier elements in the right proportions, but they all required that the processes take place in an extremely neutron-rich environment. Normal stars and supernovae utterly failed to produce the right environment. But neutron stars are almost completely made of neutrons (with a small contamination of protons and electrons), and so proponents of the neutron star merger model for short gamma ray bursts argued that this would be the right environment for the r-process.