Kilonovae, arising from neutron star mergers, produce heavy elements via the r-process and exhibit late-time infrared emission at temperatures below 1000 K, which cannot be explained by atomic transitions. Using the events AT2017gfo (GW170817) and AT2023vfi (GRB 230307A) as examples, it has been shown that conditions in kilonova ejecta allow dust formation from refractory r-process elements—zirconium, tungsten, and osmium. Kinetic calculations based on chemical reaction rate coefficients for tungsten as a representative element demonstrate efficient dust condensation, especially in slow-moving ejecta, contradicting earlier work based on classical nucleation theory. Radiation-hydrodynamic simulations including dust successfully reproduce the observed infrared emission. The formation and abundance of dust from r-process elements strongly depend on the ejecta mass, composition, and velocity. Infrared emission from such dust opens a new diagnostic method for heavy element synthesis in neutron star mergers.
Mergers of neutron stars, detected via gravitational waves (event GW170817), are accompanied by kilonovae — optical and infrared transients powered by the radioactive decay of freshly synthesized elements. Spectroscopy of such events, especially with James Webb, has allowed us to estimate the mass of ejected material and identify individual heavy elements. However, late-time observations revealed strong infrared emission with temperatures below 1000 K that could not be reproduced by models based solely on atomic lines. Back in the mid-20th century, Fred Hoyle and Margaret Burbidge showed that heavy elements are synthesized in stars, and today the detection of gravitational waves by LIGO, built with key contributions from Rainer Weiss, has opened the era of neutron-star astronomy. Kilonova spectra feature a pseudo-continuum due to Doppler broadening of numerous lines.
To test the feasibility of dust formation, researchers calculated the kinetic evolution of clusters of refractory elements (using tungsten as a representative example) in the expanding plasma of the ejecta. Unlike classical nucleation theory, they accounted for association and fragmentation reactions for clusters of all sizes, as well as the destruction of dimers by fast electrons from beta decay. Then, with numerical simulations of radiation transfer using the Sedona code, they synthesized spectra of a dusty kilonova, where the microscopic properties of dust grains (in the Rayleigh limit) set the opacity, and their temperature was computed from the heating-cooling balance incorporating Planck's radiation law. The cluster growth rate was set by molecular-dynamic coefficients for tungsten, and the formation of stable grains kicked in when the temperature dropped below ~1800 K.
It turns out that at temperatures below ~1800 K (about 10 days after the merger), dust forms very efficiently in dense, slowly expanding layers (with velocities ≲0.1c). A dust mass on the order of 10^{-3} M⊙ is enough to create an optical depth of around unity in the infrared. The outer, fast layers (≳0.2c) remain virtually dust-free due to low density. Model spectra show a nearly blackbody continuum peaking at 5 μm, which matches the spectroscopy data for AT2023vfi well. A slight excess of emission around 2.1 μm is explained by line contributions from the dust-free outer regions. For comparison, a dust-free model yields a purely line-dominated, non-thermal spectrum that looks nothing like the observations. Thus, cosmic dust from heavy elements naturally solves the mystery of late IR emission. The key takeaway: dust formation efficiency drops sharply with increasing ejecta velocity, aligning with the observed threshold of ~0.1c for both AT2017gfo and AT2023vfi.
This discovery turns the infrared glow of kilonovae into a sensitive indicator of ejecta composition, mass, and dynamics. Dust formation efficiency strongly depends on the abundance of refractory elements, which in turn is set by the pathways of nucleosynthesis. For instance, for a composition resembling the star HD 222925 with enhanced third-peak r-process, lots of dust forms, while for a composition typical of HD 122563 with depleted heavy elements, almost none does. Therefore, the observed dustiness of AT2023vfi points to a rich yield of third-peak elements, consistent with predictions from neutron star merger models. The most likely sources of slow ejecta are accretion disks and their winds, where the bulk of dust formation takes place.
Future observations, especially in the mid-infrared beyond 5 μm, will allow us to track the transition from a line-dominated spectrum to a dust continuum. Just as in supernovae of Type II we catch the moment of dust condensation, time-resolved spectroscopy of kilonovae in the first 10–20 days could directly show the birth of refractory dust grains. Moreover, the diversity of conditions in different mergers (mass, velocity, composition) should produce the full range of late-time spectra — from purely line-dominated to fully dust-dominated.
The results impact not only transient astrophysics but also galactic chemical evolution: dust formation in kilonovae could be one of the sources of interstellar refractory dust grains in ancient galaxies. It also provides a new method to gauge the production rate of heavy elements in the Universe.
Next important steps will be to account for more realistic mixtures of elements (not just tungsten) and to include dust destruction by β-particles and shock waves in kilonova remnants. Detailed non-equilibrium modeling of the ionization and heating of gas in the dust phase is also needed.
This work directly touches on the fundamental question of the origin of elements heavier than iron. If kilonovae are indeed the main 'factories' of gold, platinum, and uranium, then detecting and analyzing their dusty ejecta provides a unique tool to test theories of nucleosynthesis under extreme conditions. Moreover, dust formation links nuclear astrophysics to the physics of the interstellar medium and the processes of planet formation.
🎯 A kilonova's ejecta can synthesize hundreds of Earth masses' worth of gold, and some of it condenses into genuine cosmic dust grains that glow for a while like tiny red-hot embers.