Astrophysics · Gravitational Waves
When Neutron Stars Collide: Gold, Gravitational Waves, and the Kilonova
S
Editorial Team
September 19, 2026 · 8 min read
At 12:41:04 UTC on August 17, 2017, the LIGO and Virgo gravitational wave detectors registered a signal unlike any seen before. It lasted 100 seconds — far longer than the fraction-of-a-second chirps from merging black holes. The source was two neutron stars, each roughly 1.2 to 1.4 times the mass of the Sun, spiralling together at a distance of approximately 130 million light-years in the galaxy NGC 4993.
1.7 seconds after the gravitational wave signal ended, the Fermi and INTEGRAL gamma-ray space telescopes detected a short gamma-ray burst from the same direction: GRB 170817A. Then, 11 hours later, telescopes around the world converged on a new point of blue-white light — a transient source called AT2017gfo — fading in NGC 4993. Over the following days it shifted from blue to red, producing a spectrum rich in heavy elements.
This was GW170817 — the first astronomical event detected simultaneously in gravitational waves and light. It answered a question physicists had debated for decades: where do the universe's heaviest elements come from?
The Inspiral and Merger
The two neutron stars in the GW170817 system had been locked in a binary orbit for hundreds of millions of years, slowly losing energy to gravitational wave emission and spiralling inward. As they approached merger, the frequency and amplitude of the waves increased — the characteristic chirp signal. In the final seconds, they were completing thousands of orbits per second.
During the inspiral, tidal forces from each star deformed the other. Measuring this tidal deformability from the gravitational wave signal gives information about how stiff or soft neutron star matter is — what physicists call the equation of state. GW170817 constrained neutron star radii to roughly 11–14 km for a 1.4 solar mass star, a significant improvement on prior estimates.
At the moment of merger, the two objects became one. The remnant was briefly a hypermassive neutron star, spinning rapidly and supported against collapse by differential rotation. Within milliseconds to seconds, it likely collapsed into a black hole — the evidence from the gamma-ray burst suggests a relativistic jet was launched, which typically requires a compact remnant.
"The gold in your jewellery, the platinum in your catalytic converter, the iodine in your body — these atoms were forged in a neutron star collision somewhere in our Galaxy's past. GW170817 was not just an astronomical event. It was a direct view of the forge."
Enjoying this? Get one article like this every week — free. Subscribe below →
The Kilonova: A Factory of Heavy Elements
When neutron stars merge, they eject a fraction of their mass — typically 0.01 to 0.05 solar masses — at velocities of 10–30% the speed of light. This neutron-rich ejecta is the site of the rapid neutron capture process (r-process): the most efficient way nature has to build elements heavier than iron.
In the r-process, atomic nuclei are bombarded by neutrons faster than they can undergo radioactive decay. Nuclei rapidly accumulate neutrons, climbing the chart of nuclides to masses far beyond what stellar fusion can reach, then decay back toward stability — producing the heavy elements. Neutron star mergers provide exactly the right environment: extreme neutron flux, extreme temperature, and neutron-rich matter.
The transient AT2017gfo confirmed this. Its early blue emission came from lighter r-process elements (strontium was directly identified, with an abundance 5–10 times the mass of Earth). Its later red emission — the ejecta glowing in infrared as lanthanides and actinides formed — was the signature of heavier r-process products. The opacity of lanthanide-rich matter reddens and dims the emission over days, producing the characteristic colour evolution a kilonova displays.
Theoretical models suggest a single neutron star merger can produce:
- Roughly 10 Earth masses of gold
- Similar quantities of platinum, uranium, iodine, and dozens of other r-process elements
- Strontium, barium, and europium — elements long associated with peculiar stars but now understood to come from mergers
The rate of neutron star mergers in the Milky Way — estimated at roughly one every 10,000–100,000 years — when integrated over the Galaxy's history, is broadly consistent with the observed cosmic abundances of r-process elements. The merger hypothesis, long theoretical, now has direct observational support.
The Gamma-Ray Burst Connection
Short gamma-ray bursts (SGRBs) — intense flashes of gamma rays lasting less than two seconds — had been theorised since the 1980s to originate from neutron star mergers. GW170817/GRB 170817A confirmed this directly. But GRB 170817A was unusually faint for its distance, suggesting we were not looking straight down the jet.
The jet was seen off-axis — at roughly 15–30 degrees from the line of sight. Over the following months and years, radio and X-ray observations tracked the afterglow as the jet swept up surrounding interstellar gas and decelerated. The morphology was consistent with a structured jet viewed from the side: a narrow ultra-relativistic core with a surrounding cocoon of slower ejecta. This was the first direct structural observation of a gamma-ray burst jet.
What We Still Don't Know
GW170817 answered many questions but raised others. The fate of the merger remnant is still debated — whether the hypermassive neutron star survived for a long or short time before collapsing into a black hole affects the outflow and the r-process yield, and the gravitational wave data from the post-merger phase (if any signal was emitted above detector sensitivity) is ambiguous.
The ratio of neutron star mergers to core-collapse supernovae as r-process sites is uncertain. Some metal-poor stars in the halo of our Galaxy show r-process enhancement at very early cosmic times, which may be hard to explain if mergers were the only source — because binary neutron star systems take time to form and inspiral. Magnetar-driven jets from rare supernovae may be an additional r-process site. The story is not yet complete.
LIGO, Virgo, and KAGRA continue to improve sensitivity with each observing run. The next neutron star merger within detection range — expected within years — will be studied with more telescopes, faster response, and better prepared models. Each event is another reading from the forge.
What Does the Merger Leave Behind?
When two neutron stars merge, what survives? This depends critically on the total mass of the system and the neutron star equation of state — properties that are still being constrained.
The three possible outcomes
- Prompt collapse to a black hole: If the combined mass exceeds the maximum mass that even a differentially rotating neutron star can support (roughly 2.5–3 solar masses for most equations of state), the merger remnant collapses to a black hole within milliseconds. The gravitational wave signal would show a merger chirp followed by an abrupt cutoff.
- Hypermassive neutron star: The remnant exceeds the maximum mass for a cold, rigidly rotating neutron star but is temporarily supported by differential rotation and thermal pressure. It survives for tens to hundreds of milliseconds before collapsing to a black hole. During this time it radiates gravitational waves at kilohertz frequencies — potentially detectable with next-generation detectors — and drives intense magnetic fields that may power short gamma-ray bursts.
- Long-lived or stable remnant: For sufficiently stiff equations of state (ones where neutron star matter is relatively incompressible), the remnant might survive for seconds, years, or indefinitely. A stable remnant would be a new massive millisecond pulsar, spinning rapidly and losing energy to magnetic dipole radiation. Evidence for a long-lived remnant has been searched for in GW170817 but not found — the data are consistent with a collapse within ~1 second.
The gravitational wave signal in the post-merger phase — above roughly 1 kHz — carries the most direct information about the remnant and the equation of state. Current LIGO/Virgo sensitivity is insufficient to resolve this for all but the closest events. The Einstein Telescope (planned for Europe) and Cosmic Explorer (planned for the US), both envisioned for the 2030s, will have sensitivity in this frequency range and may hear dozens of post-merger signals per year.
The Bigger Picture: Nucleosynthesis and the Origin of Heavy Elements
GW170817 confirmed that neutron star mergers are a significant site of r-process nucleosynthesis — but are they the only site? This question has implications stretching across stellar evolution, galaxy chemical evolution, and the history of the elements.
The r-process problem
The rapid neutron capture process (r-process) requires extreme neutron fluxes found only in a few environments: core-collapse supernovae (specifically the neutrino-driven wind above the proto-neutron star) and neutron star mergers. For decades, before GW170817, the r-process site was uncertain. The kilonova observation shifted the consensus toward neutron star mergers as the dominant site, but several lines of evidence complicate a simple picture:
- Chemical evolution of early galaxies: Metal-poor stars in the Milky Way's halo show r-process enhancements that appear too early in galactic history for neutron star mergers — which require time for the two neutron stars to form, inspiral, and merge — to account for. Core-collapse supernovae, which occur within millions of years of star formation, are needed to explain the early enrichment.
- Rare but prolific events: The r-process abundance patterns in ancient stars suggest a rare, prolific source. Neutron star mergers are rare (perhaps 1 per 10,000 supernovae in a typical galaxy) but each one produces a large mass of r-process material — consistent with this pattern.
- Collapsars: Some models propose that rapidly rotating massive stars that collapse to form long gamma-ray bursts (collapsars) can also drive r-process nucleosynthesis in their accretion discs and jets. If true, this would add another site to the inventory.
The current consensus is that neutron star mergers and possibly collapsars are the dominant r-process sites, with core-collapse supernovae contributing a lighter subset of r-process elements. The debate continues, and each new gravitational wave event with an associated kilonova observation adds a new data point to constrain the models.
The gold in your wedding ring
The practical implication is personal: every atom of gold, platinum, iridium, and uranium on Earth was synthesised in a cataclysmic merger or explosion billions of years before the Sun formed, scattered through space, and eventually incorporated into the molecular cloud that became our solar system. Gold's cosmic abundance is not large — roughly 0.003 parts per million in the Sun by mass — precisely because the events that make it are rare. That rarity is encoded in the weight of precious metals in your hand.
