On July 28, 2019, the LIGO and Virgo gravitational-wave observatories recorded a brief shudder in spacetime: two black holes, roughly 12 and 8 times the mass of the Sun, had spiraled together and merged about 2.7 billion light-years from Earth. The event was cataloged as GW190728 and, at first glance, looked like dozens of other mergers already on the books. Five years later, a team led by MIT physicist Kaze Wong has found something unusual hiding in that signal, something that could mark the first time dark matter left a measurable imprint on a gravitational wave.
Their study, posted as a preprint in late 2025 and referenced in what appears to be an MIT institutional summary (the exact URL and publication date have not been independently verified), reanalyzed 28 well-characterized black hole mergers from the LIGO-Virgo-KAGRA (LVK) catalog. Twenty-seven of them looked exactly as expected: two black holes colliding in empty space. GW190728 did not. When the researchers allowed for a surrounding cloud of ultralight scalar field, a leading theoretical candidate for dark matter, the data fit better than the standard vacuum model. The team described the result as showing “a preference for the dark-matter model,” language chosen to signal a tantalizing hint rather than a confirmed discovery.
Why dark matter might show up in gravitational waves
Dark matter accounts for roughly 27 percent of the universe’s total energy budget, according to measurements by the European Space Agency’s Planck satellite. It outweighs all visible stars, gas, and dust by a factor of about five. Yet despite decades of searching with underground particle detectors, orbiting telescopes, and particle colliders, no experiment has caught dark matter interacting with ordinary matter in a laboratory. Every detection so far has been indirect, inferred from the gravitational pull dark matter exerts on galaxies and light.
Gravitational waves offer a different approach entirely. Instead of waiting for a dark matter particle to bump into a sensor, physicists can look for the collective gravitational effect of a dark matter cloud on the motion of massive objects, specifically black holes spiraling toward collision.
The theoretical groundwork for this idea has been building for years. One class of dark matter candidates, ultralight scalar fields, would behave more like a wave than a particle at astrophysical scales. Around a rapidly spinning black hole, a quantum process called superradiance can amplify these waves, pulling energy from the black hole’s rotation and building a dense fog of scalar field energy in its immediate neighborhood. When a second black hole plunges through that fog, the extra mass-energy environment changes the gravitational-wave signal in specific, calculable ways.
Research on dynamical friction in ultralight dark matter showed that a wave-like medium exerts drag on moving black holes, slowing their inspiral and shifting the phase evolution of the emitted gravitational waves. Separate numerical simulations validated analytic expressions for relativistic drag forces from scalar clouds, confirming that the effect is large enough, in principle, to register in current detector data. Those upstream results do not depend on the GW190728 finding and would stand even if this particular hint fades.
What the analysis actually found
Wong’s team built a waveform model designed for compact binaries embedded in a light scalar field environment, then ran a Bayesian comparison against every clear merger in the LVK catalog. Bayesian analysis is a statistical method that weighs how well competing models explain the same data while penalizing models that use extra free parameters. It is a standard tool in gravitational-wave astronomy, used precisely because it guards against the temptation to over-fit noisy signals.
For 27 of the 28 events, the vacuum model and the scalar-field model performed equally well, meaning there was no evidence of anything unusual. GW190728 broke the pattern. The scalar-field model was statistically preferred, suggesting the waveform carried subtle phase shifts consistent with a surrounding dark matter cloud.
That does not mean dark matter has been detected. A single anomalous event in a catalog of 28 could easily be a statistical fluctuation. The more complex model introduces additional parameters (the scalar field’s mass, its density profile, the spin history of the progenitor black holes), and even with Bayesian penalties, extra parameters can sometimes latch onto noise. The researchers acknowledged this openly, framing GW190728 as tentative evidence rather than a detection.
The hurdles still standing
Several significant uncertainties surround the result. The full posterior distributions and prior choices for the scalar-field parameters in GW190728 have been summarized in the preprint but not released as open data tables or reproducible code. Until an independent group re-implements the waveform model against raw LVK strain data, the result cannot be externally confirmed.
The LVK collaboration itself has not issued a public statement on the environmental interpretation of GW190728. That silence is routine: the study is a preprint, not yet peer-reviewed, and the collaboration typically reserves formal commentary for published results that clear its internal review process.
Model dependence is another concern. The density and structure of a scalar cloud around a binary black hole hinge on assumptions about the scalar field’s mass and self-interaction strength. Strong self-coupling, for instance, can smooth out the very density spikes that make the effect detectable, raising the bar for how dense the dark matter environment must be to leave a visible trace. Different parameter choices can produce dramatically different predictions for the waveform deviation.
Instrumental and astrophysical systematics add further complexity. Gravitational-wave detectors are extraordinarily sensitive but also noisy, with artifacts that can sometimes mimic or distort real signals. Ordinary matter (gas disks, stellar remnants, or dense star clusters in a galactic nucleus) can also perturb a binary’s orbit. The current analysis argues that the phase shifts in GW190728 match scalar-field expectations better than standard astrophysical environments, but a comprehensive comparison against every plausible mundane explanation has not yet appeared in the literature.
What comes next
The practical significance of the result may lie less in GW190728 itself than in what it proves about the search method. Current detectors are already sensitive enough to begin probing dark matter environments around merging black holes. That is a meaningful milestone regardless of whether this particular event holds up.
LIGO, Virgo, and KAGRA are preparing for their next major observing campaign, known as O5, expected to begin in 2027 with upgraded sensitivity that could roughly double the volume of space the detectors survey. Hundreds of new mergers should enter the catalog. If ultralight scalar dark matter clouds are real and common enough, patterns should emerge: repeated hints at similar scalar-field parameters across multiple events would dramatically strengthen the case. An absence of corroborating signals would push the community toward more conservative explanations for GW190728.
For now, the evidence supports a measured reading. Theoretical work has established that light scalar dark matter could leave detectable fingerprints on gravitational waves. A careful new analysis has identified one event where such a fingerprint might be present. The data are not yet strong enough, and the modeling not yet mature enough, to declare that dark matter has been seen. But a new observational window has cracked open. With time, more mergers, and independent scrutiny, it could reshape how physicists hunt for the invisible mass that holds galaxies together.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
