A 0.4-Second Gamma-Ray Burst Had a Nine-Minute X-Ray Secret

On July 4, 2025, at 08:16:27 UTC, three gamma-ray monitors—SVOM/GRM, Insight-HXMT, and Konus-Wind—caught a brief flash from deep space. The burst, named GRB 250704B, lasted about 0.37 seconds. That duration is a classic signature of a compact object merger, the violent coalescence of neutron stars or a neutron star with a black hole.

If gamma rays had been the only available window, the story would have ended there. But the Einstein Probe's Wide-field X-ray Telescope was watching the same region in softer X-rays—lower-energy radiation than gamma rays—and it saw something far longer. Starting at T0 + 0.2 seconds, the X-ray source EP250704a appeared. It began with a sharp spike lasting about 0.35 seconds, followed by a tail and then a broad bump that persisted for roughly 540 seconds. In total, the X-ray emission remained detectable for about 560 seconds.

A short burst with a long X-ray shadow

Short gamma-ray bursts are thought to arise when compact objects merge, launching a relativistic jet. For years, astronomers have debated whether some merger-driven bursts also produce long-lasting X-ray emission during the prompt phase—the period when the central engine is still active—or whether such X-rays are simply the beginning of the afterglow, the later glow produced when the jet slams into surrounding gas.

The new event makes that distinction concrete. The sharp X-ray spike lines up with the second gamma-ray pulse of GRB 250704B, linking the two signals to the same explosion. But during the minutes-long X-ray bump, the gamma-ray monitors detected nothing down to a limit of 2.13 × 10^-9 erg cm^-2 s^-1 in the 15–150 keV band. The X-ray phase was bright enough for Einstein Probe yet quiet in gamma rays.

Why the X-rays point inward, not outward

Afterglow emission is usually smooth and predictable, because it comes from an external shock plowing into the surrounding medium. The long X-ray bump did not behave that way. The authors report rapid variability, with changes occurring on timescales short compared with the emission's age. They also find a spectral hardening around T0 + 350 seconds, interpreted as the handoff from fading prompt emission to the emerging afterglow. Earlier, the X-ray behavior resembles the early plateaus seen in some short bursts, which are often attributed to a central engine.

Those clues point inward. The paper concludes that the extended X-ray component up to about T0 + 400 seconds is engine-powered, not simply afterglow. Later observations strengthen the case: the X-ray afterglow shows a shallow decay from T0 + 690 seconds to about 7.2 × 10^4 seconds, while the optical light rises slowly from T0 + 2 × 10^3 seconds to 6.5 × 10^4 seconds. The authors argue that this requires sustained energy injection from a long-lived central engine—most likely a magnetar, a rapidly spinning, highly magnetized neutron star left behind by the merger.

Inferring a merger without seeing the merger

The team established the burst's distance using the Very Large Telescope's X-shooter spectrograph. Absorption lines in the afterglow light place it at a redshift of 0.66102 ± 0.00011, meaning cosmic expansion has stretched its light during its long journey. At that distance, the prompt gamma-ray spectrum peaks at about 978 keV in the burst's rest frame, and the isotropic-equivalent energy is about 3.79 × 10^51 erg. Those values place GRB 250704B among merger-driven bursts, or Type I, rather than bursts from collapsing massive stars, or collapsars.

Deep optical limits also rule out a bright supernova. That matters because a collapsing massive star would likely leave such a signature, while a compact merger would not. At this redshift, however, the unmistakable merger signals—a gravitational-wave chirp or an optical/infrared kilonova—are beyond the reach of current ground-based facilities. The merger identification therefore rests on the burst's gamma-ray properties and the absence of a supernova, not on a direct gravitational-wave detection.

How many more are hidden?

If astronomers had monitored GRB 250704B only in gamma rays, they would have seen only the 0.4-second spike. The paper estimates that extrapolating the average flux of the X-ray bump into the 15–150 keV band would put it below the detection threshold of Swift's Burst Alert Telescope. Swift/BAT detection becomes feasible at z ≲ 0.08, and even then the bump would be at best marginal. Yet short gamma-ray bursts localized by Swift typically sit at redshifts greater than about 0.1, well beyond the distance where such a faint gamma-ray excess could be seen.

That raises a provocative possibility: the long X-ray phase may be common in merger-driven bursts, hidden from most gamma-ray telescopes. If so, wide-field X-ray monitors could provide a valuable electromagnetic counterpart to gravitational-wave sources, catching the engine's long goodbye after the gamma-ray flash.

What one event can and cannot prove

This is a single event, and the work appears in a preprint that has not yet completed peer review. The claim that long-lasting X-ray emission is a common feature of merger-driven bursts is an extrapolation from one well-observed case, not a proven population statistic. The internal-engine interpretation is also a model-based conclusion, though it is supported by rapid variability, spectral evolution, and late-time energy requirements.

Still, the observation changes the picture. A burst that looked like an ordinary short gamma-ray burst in one band turned out to have a nine-minute X-ray secret in another. The next step is to see whether other mergers whisper for minutes in X-rays while staying nearly silent in gamma rays—and whether gravitational-wave observatories can help catch them in the act.