A growing body of evidence indicates that binary neutron-star mergers are the primary origin of heavy elements produced exclusively through rapid neutron capture1,2,3,4 (the ‘r-process’). As neutron-star mergers occur infrequently, their deposition of radioactive isotopes into the pre-solar nebula could have been dominated by a few nearby events. Although short-lived r-process isotopes—with half-lives shorter than 100 million years—are no longer present in the Solar System, their abundances in the early Solar System are known because their daughter products were preserved in high-temperature condensates found in meteorites5. Here we report that abundances of short-lived r-process isotopes in the early Solar System point to their origin in neutron-star mergers, and indicate substantial deposition by a single nearby merger event. By comparing numerical simulations with the early Solar System abundance ratios of actinides produced exclusively through the r-process, we constrain the rate of occurrence of their Galactic production sites to within about 1?100 per million years. This is consistent with observational estimates of neutron-star merger rates6,7,8, but rules out supernovae and stellar sources.

As neutron-star mergers are rare—they only occur a few times per million years (Myr) in the Milky Way9—their production of heavy r-process elements is infrequent. This results in the strong temporal variation of the abundance of r-process elements in the interstellar medium. Other plausible r-process sources, mainly supernovae and massive stars, are orders of magnitude more frequent in the Milky Way, such that r-process production by them can be approximated as being uniform in time5,10 (the ‘uniform production’ model)...

...We used short-lived r-process isotopes to constrain the Galactic rate of occurrence (hereafter ‘rate’) of r-process production sites. These elements encode the short-term history of their production, making them a sensitive indicator of the rate of their source. Several such elements have measured abundances in the early Solar System5. These include two actinides, 247Cm (with half-life t1/2 = 15.6 Myr) and 244Pu (t1/2 = 80.8 Myr), and 129I (t1/2 = 15.7 Myr). For comparison we additionally examined 235U (t1/2 = 703.8 Myr), a radioactive actinide with longer half-life and known abundance in the early Solar System.

We used our simulations to derive the probability distribution of (N247Cm/N244Pu)ESS for different Rmerger values. We computed the density of actinides in the interstellar medium at the location of the pre-solar nebula, about 8.3 kpc from the Galactic Centre, for a range of Galactic ages from 8,500 to 9,500 Myr. Figure 2 shows the simulated interstellar-medium abundance ratio (N247Cm/N244Pu)ISM,sim as a function of time for one Monte Carlo realization, for two different merger rates. For the higher rate shown, Rmerger = 500 Myr?1, we see that (N247Cm/N244Pu)ISM,sim is distributed closer to the abundance ratio predicted by the uniform production model than is the curve we calculate for the lower merger rate, Rmerger = 20 Myr?1 (see below). This is expected, because the uniform production model is approximately the same as an infinite source rate. This higher rate of Rmerger = 500 Myr?1, while still much lower than the Galactic core-collapse supernova rate15 of about 3 × 104 Myr?1, is inconsistent with the early Solar System abundance ratio.

When neutron stars merge, they create an accreting black hole (the accretion disk is shown red). Tidal (dynamical) forces and winds from the accretion disk eject neutron-rich matter. This ejected matter (ejecta, shown grey) undergoes rapid neutron capture, producing heavy r-process elements, including actinides. The ejecta reach the pre-solar nebula and inject the heavy elements that will remain in the Solar System.

Values are shown as functions of time measured from the formation of the Milky Way. a, The abundance ratio (N247Cm/N244Pu)ISM,sim in the interstellar medium near the pre-solar nebula for Galactic merger rates of 20 Myr?1 (blue line) and 500 Myr?1 (grey dashed line), for the early Solar System abundance ratio (N247Cm/N244Pu)ESS (ESS; red line) and for the abundance ratio predicted by the uniform production model (UP; green line), the last two both shown with 1? error regions. Also shown is the 90% confidence interval of (N247Cm/N244Pu)ISM,sim for Rmerger = 20 Myr?1 (blue shading). b, Distance of the merger with the greatest contribution of curium to the early Solar System (rdom), assuming Rmerger = 20 Myr?1. c, Fraction of curium in the early Solar System from the single dominant source, assuming Rmerger = 20 Myr?1.

a–d, As for Fig. 2a but for abundance ratios (N247Cm/N232Th)ISM,sim (a), (N129I/N127I)ISM,sim (b), (N244Pu/N232Th)ISM,sim (c) and (N235U/N238U)ISM,sim (d).

Our computations show that the dominant source contributed a substantial part of (N247Cm)ESS. The single merger with the highest contribution deposited fdom,Cm = 70% ± 20% of the 247Cm. Because all other isotopes of curium have a much shorter half-life than 247Cm, this means that 70% ± 20% of all curium in the early Solar System was produced by a single neutron-star merger.

We carried out the same calculation for 244Pu. Owing to its longer half-life, the early Solar System abundance (N244Pu)ESS is dominated to a smaller extent by a single event than (N247Cm)ESS. We found that a single source deposited fdom,Pu = 40% ± 15% of the 244Pu. Because all other isotopes of plutonium have a much shorter half-life than 244Pu, this means that 40% ± 15% of all plutonium in the early Solar System was produced by a single neutron-star merger. Interestingly, the dominant source of plutonium is not always the same merger as the dominant curium source.