The TDT concept

The spectroscopic bottleneck

Time-domain surveys from the ground and in space, together with gravitational-wave and neutrino detectors, produce hundreds of alerts every night. Yet today fewer than 10% of transients are identified spectroscopically, and fewer than 1% are studied in detail. The bottleneck is the limited number of target-hours on the world’s optical and near-infrared telescopes: time-domain science needs repeated observations, its targets are sparse (typically fewer than one per square degree per night), and the spectra are often time-critical. The shock break-out of a supernova must be caught within minutes to hours of the explosion; a transmission spectrum must be taken during the planet’s transit.

The TDT addresses this bottleneck by combining extreme adaptive optics, near-read-noise-free detectors and replication with flexible, robotic scheduling. It probes far below the optical sky background that limits current facilities for point sources, while providing the multiplicity to monitor many events scattered across the sky.

At a glance

The telescope array

The TDT does not push the aperture of a single telescope. It replicates a technically modest but operationally powerful unit about a hundred times and coordinates the copies. Each unit is robotic and permanently instrumented, so there are no instrument changes, and failures or maintenance on individual units never take the facility down.

The units are individually steerable, so users can go big or go wide: all units on one target for maximum depth, every unit on a different target for maximum multiplex, or any combination of sub-arrays in between. Light from units on the same target is combined at the data level, so observing power grows with the number of units.

On first analysis, a 2 m aperture is the sweet spot between cost, manufacturability, the aperture needed for adaptive optics in the blue, and the spatial resolution needed to resolve out the sky background. Smaller units would add maintenance overhead and not suppress the sky enough; larger units would cost substantially more, need many more adaptive-optics actuators, and reduce how many units can share a target.

Extreme adaptive optics

Each unit corrects the atmosphere with extreme adaptive optics in the optical blue, aiming for at least 80% enclosed energy within about 0.12″ at 400 nm. Compared with spectrographs designed for 1″ seeing, this dilutes the sky background by a factor of 100, separates sources in crowded fields, and makes the instruments far more compact. More about extreme adaptive optics

Fibres and spectrographs

The adaptive-optics-corrected focal plane of each unit is sampled by a seven-fibre bundle, with fibres of about 0.12″ in the blue and larger in the red, where the diffraction limit is larger. The fibres of a cluster of units are arranged into a pseudo-slit that feeds a shared set of spectrographs, which together cover:

• spectral resolutions from R = 10³ to 10⁵;
• 350–1800 nm (u to H band) simultaneously up to R ≈ 5000, similar to X-Shooter, with higher resolutions in selected wavelength ranges.

Sharing a spectrograph spreads the light of many units over many detector pixels. Near-read-noise-free detectors, below 0.1 e⁻ per pixel and with dedicated rows for each unit, remove the read-noise penalty and let units start and stop independently. At lower resolutions, photonic lanterns or spectrographs on a chip may allow one spectrograph per unit. More about the detectors

Performance

From version 1.0 of the TDT exposure time calculator:

• With all 100 units on one target, the TDT reaches g ≈ 25 in 30 minutes at a signal-to-noise ratio of about 10 per resolution element at R = 5000, matching spectroscopically the imaging depth of the Vera C. Rubin Observatory, Euclid and Roman.
• In a 10-minute exposure at R = 1000, a signal-to-noise ratio of 10 is reached at g ≈ 20.5, 23.0 and 25.3 with 1, 10 and 100 units.
• At R = 10⁵, the full array reaches a signal-to-noise ratio of about 200 per resolution element in one hour on a g = 16 star.

¹ From the ESO exposure time calculator. ² The ELT mode closest to the TDT; ELT adaptive optics will not be available in the optical blue.

A single TDT unit is comparable to VLT + X-Shooter in the sky-background-limited regime, and with laser tomography adaptive optics the ELT performs like a 30-unit TDT, so the point-source survey speed of the full TDT is three times that of the ELT. Full performance comparison, including the WST

Operations

The TDT is a PI-driven facility, like the VLT: proposals are ranked by a time allocation committee, and users ask for unit-hours, which can be used in any configuration of telescopes, targets, spectral resolution and cadence. Large and multi-year programmes may be offered, but the TDT is not just a survey instrument.

Effectively every observation is a hard target of opportunity, so the whole array must be able to reconfigure within a minute. An agentic approach, with AI agents for each telescope, for scheduling, for data reduction and for maintenance, manages the array within strictly defined optimisation parameters. More about automation and operations

A global array

Because time-domain spectroscopy is so often time-critical, the TDT is planned as a global facility. Six stations at existing observatories, as for networks such as Las Cumbres Observatory, can reach any position on the sky at any time, with a redundancy factor of 2.4 against weather and technical downtime. A three-station array in the Southern Hemisphere, the bare minimum, would cover 70% of the sky around the clock, with a redundancy factor of 1.2.

The core at ESO would be the full configuration of about 100 units; the other stations could be smaller clusters that add time coverage at the cost of depth, built under the Expanding Horizons programme or in partnership with other communities. To avoid laser guide star interference and spread the weather risk, the core itself may be distributed over several sites, for example Cerro Paranal, Cerro Armazones, and La Silla with its extension Cerro Vizcachas.

Roadmap

The TDT is being developed for ESO’s Expanding Horizons programme, which seeks a transformational facility for the 2040s. A first version, proto-TDT/FAROS, will adapt the four Auxiliary Telescopes of the VLT Interferometer at Paranal to test the robotic adaptive optics, the detectors and the scheduling together. The modular design allows a gradual roll-out, with science from the prototype and the first units almost from day one, and the required technologies within reach on a 15-year timescale. See the development path and the timeline.