A gravitational-wave detector catches not only the roar of mergers but also the quiet splash when a star or gas cloud dives into a supermassive black hole. Such a dive, like a stone falling into a well, generates waves at all frequencies at once—and in that spectrum lies encoded the nature of the well itself. If at the bottom, instead of a bottomless pit, there lies a reflecting plate, the echo returns differently.

Physicists have long suspected that not all black hole candidates possess a true event horizon. Instead, there might be a reflecting surface at a distance rs=2M(1+ε), where ε is a tiny addition, down to the Planck length. Such objects are called black hole mimickers. Until now, they have been sought via resonances in the inspiral phase or via echo signals after merger, but both methods are limited to frequencies around the last stable orbit. The new work proposes a different tool: the spectrum of the plunge itself, which looks beyond that threshold.

The authors built a surrogate model: instead of a full merger, they consider the direct fall of a light particle onto a massive object with a reflecting surface. Metric perturbations are described by the Regge–Wheeler and Zerilli equations, and the energy spectrum is computed via the Green's function. Calculations for modes (ℓ,m)=(2,2) with different reflection coefficients R (from 1 to 10⁻³) and parameters ε (from 10⁻⁴ to 10⁻¹⁰) revealed two robust signatures. At low frequencies, a comb of resonances appears with spacing δω=π/(2M|log ε|). The main surprise awaits above the threshold Mωth≈0.39: for a black hole the spectrum decays exponentially, while for a mimicker an additional contribution arises, proportional to |R|². At R=1 and Mω≈0.7, the energy exceeds the black-hole case by orders of magnitude, and this excess is almost independent of ε and the particle's angular momentum.

Such a high-frequency excess resembles an echo sounder that listens not to the reflected signal but to the splash itself upon impact. It is robust to details of the source and the position of the surface, making it a reliable signature of the absence of a horizon. While a single plunge in an extreme-mass-ratio inspiral yields a low signal-to-noise ratio, coherent stacking of many events could make the signature detectable. LISA, planned for launch in the 2030s, could accumulate enough such plunges from stars falling onto supermassive black holes in galactic centers.

The next step is a full time-domain analysis of the plunge–merger–ringdown waveform for mimickers, with proper treatment of the source after it crosses the surface. This will require careful windowing methods to avoid artifacts from source truncation. Another direction is generalization to plunges from the last stable orbit and to rotating objects. If future data show a high-frequency excess, it would be the first observational evidence that the event horizon is not the end of physics but a door.