The planned Einstein Telescope would still deliver most of its promised science even if parts of its design fell short of baseline expectations, according to a preprint from the collaboration behind the proposed observatory. The study, posted on arXiv on July 29, systematically degrades the detector’s projected noise performance and tracks the losses across a wide range of science goals, from routine merger detections to elusive signals like supernova bursts.
The Einstein Telescope is a proposed third-generation gravitational wave observatory, larger and more sensitive than today’s LIGO, Virgo, and KAGRA detectors, which together have logged roughly 400 confident detections to date. Two geometries are under active consideration: a triangular design with three interferometers and 10 kilometer (6.2 mile) arms, and a configuration of two L-shaped interferometers with 15 kilometer arms at separate European sites. Each consists of a cryogenic low-frequency instrument covering roughly 3 to 30 hertz and a room-temperature high-frequency instrument reaching several kilohertz.
The paper, led by Ulyana Dupletsa and colleagues across European institutions, builds a framework for comparing how changes in specific noise sources translate into scientific losses. It considers four representative design parameters: the length of the filter cavity in the low-frequency instrument, the coating and suspension temperatures, and the beam size in the high-frequency instrument. Each is varied from a baseline value through an intermediate case and a worst case, such as raising the cryogenic temperature from 10 kelvin to 20 kelvin and then 70 kelvin, or shortening the filter cavity from 5 kilometers to 1 kilometer. The authors stress that the intermediate and worst cases are defined solely for the study and are not intended as realistic projections of what the detector will achieve.
The framework also includes a “traffic light” analysis that degrades sensitivity by a factor of 1.5 across five frequency bands, independent of which noise source causes the change. The aim is a diagnostic tool: given a proposed loss of sensitivity, scientists can see which science cases suffer most and which frequency ranges matter for what.
The headline result is that the science case survives even the most severe degradations considered. In the worst-case scenario, in which all four parameters are degraded together, detection rates remain high: about 10,000 binary neutron star mergers and 48,000 binary black hole mergers per year, compared with baselines of roughly 44,800 to 78,600 neutron star mergers and 77,900 to 91,400 black hole mergers depending on geometry. The paper notes the current generation detects about 150 compact binary mergers per year.
The losses are not spread evenly. Sensitivity below 30 hertz governs the detectability of massive and high-redshift black hole binaries, early-warning alerts for neutron star mergers, and searches for the stochastic gravitational wave background. A degraded low-frequency instrument can cut the detection horizon by as much as 40 percent over much of the mass range, and the number of very high signal-to-noise events, the ones used for precision tests of general relativity, falls by roughly a factor of five. The band from 30 to 450 hertz controls most parameter estimation for compact binaries. Above 450 hertz, the impact is concentrated on neutron star post-merger signals, tidal deformability measurements, and core-collapse supernova detection.
Among individual parameters, the suspension temperature had the largest effect, shifting detection horizons by more than 20 percent across the mass range in the worst case. Intermediate-case degradations produced only minor changes across most metrics. The paper also finds the two-interferometer configuration generally more robust than the triangle for neutron star science, driven mainly by its longer arms, though the authors caution that a direct comparison of the two geometries falls outside the study’s scope.
The authors are explicit about what the work does not establish. The intermediate and worst-case scenarios are not forecasts of detector performance, and the results are framed as comparative diagnostics rather than absolute predictions. Population-based metrics carry astrophysical uncertainties, the Fisher-matrix approach used for parameter estimation has known limitations, and the supernova results assume perfect signal recovery, which the authors call optimistic. Correlated noise between detectors, which could hurt stochastic background searches below roughly 40 hertz, is not included.
The preprint appears as the Einstein Telescope project moves toward implementation, with candidate sites in Sardinia, the Belgian-Dutch-German border region, and Lusatia in Germany, according to the project’s website. The study was carried out as part of the Einstein Telescope Organization’s design process, and the authors describe the goal not as optimizing the detector but as understanding design trade-offs before the final configuration is fixed. Their conclusion is that the observatory’s core science case is robust: even if first-stage sensitivity falls short of the target, the telescope should still detect binaries at cosmological distances, measure their parameters precisely, and probe sources no current detector can reach.
Sources
1. arXiv, “Assessing the Impact of Instrumental Requirements on the Scientific Performance of the Einstein Telescope”: https://arxiv.org/abs/2607.27311
2. Einstein Telescope Project, “Home”: https://www.et-gw.eu/