Can One Giant Telescope Act Like 33? Inside the ELTI Concept for the 2040s

In a white paper prepared for the European Southern Observatory's call for white papers about astronomy in the 2040s, Francisco Prada of the Instituto de Astrofísica de Andalucía (CSIC) and colleagues propose a new mode for the Extremely Large Telescope. They call it ELTI: the Extremely Large Telescope Interferometer. The concept is posted to arXiv. It uses the telescope's own segmented primary mirror as a compact interferometer: the ELT's M1 segments are grouped into sub-pupils, each of which forms an image on a fast single-photon detector.

A 33-Eye Interferometer in the ELT's Mirror

The ELT's primary mirror, M1, is made of 798 segments. In the ELTI layout, those segments are grouped into 33 sub-pupils, each 7.25 meters across. Nineteen segments form one sub-pupil. The arrangement uses 627 segments—79 percent of the primary—and leaves the rest unused. The centers of the 33 sub-pupils define 528 independent baselines, the pair separations that determine how fine a detail the interferometer can resolve. The longest baseline is 33.5 meters.

Each sub-pupil would form an image on a detector. In the proposed example, a 2,000×2,000-pixel SPAD camera—short for single-photon avalanche diode—records 33 spots in a fixed pattern. The spacing is designed to keep 90 percent of the light from each spot within a 0.5-arcsecond atmospheric blur at 4500 Å. For a uniformly illuminated disk 2.75 milliarcseconds—thousandths of an arcsecond—across at 4500 Å, the paper shows a simulated visibility pattern. In effect, the ELT would behave like 33 telescopes of 7.25 meters each, separated by up to 33.5 meters.

Picosecond Timing for Photon Correlations

Intensity interferometry measures how brightness fluctuations at two separated sub-pupils correlate in time. The paper's version uses the ELT's sub-pupils as the collectors and relies on extremely fast photon timing. SPAD arrays are suited to that job. They can register individual photons with timing resolutions of a few tens of picoseconds—trillionths of a second—and with low dark counts and high detection efficiency across the optical band. Until recently, astronomical SPADs were mostly single-pixel devices, which limited them to measuring one point. Two-dimensional arrays are now changing that. The paper points to a 64×64-pixel SPAD sensor developed by the Instituto de Microelectrónica de Sevilla (CSIC-US) and its first astronomical results.

From a V=6 Star to Milliarcsecond Science

In a single narrow wavelength channel—4500 Å with a width of 0.1 Å—the paper estimates that the 528 baselines could reach a signal-to-noise ratio of 5 for a V=6 magnitude star in about 100 seconds, assuming an overall instrumental efficiency of 0.5 and 10-picosecond time sampling. That is bright-star territory. To go fainter, the concept adds a high-resolution spectrograph that disperses light into roughly 50,000 spectral channels between 3500 and 9000 Å. Because signal-to-noise grows with the square root of the number of independent channels, the multi-wavelength mode could reach targets as faint as g-SDSS ≈14 magnitude in one hour.

The paper presents those numbers as expected performance for a design concept: the required megapixel SPAD detector is under competitive procurement, while the spectrograph and data-analysis approach are discussed as part of the proposal. The paper is a design concept for the 2040s, not a report of an observation.

If it works, the payoff would be angular resolution and spectral information at visible wavelengths. The paper says ELTI could enable true milliarcsecond imaging of stellar surfaces, direct probes of extreme-gravity and strong electromagnetic-field environments around compact objects, precision studies of accretion physics, and even the potential detection of Earth-sized exoplanets through minute visibility modulations. The exoplanet case is explicitly a potential, not a demonstrated result. The same is true of the broader science goals: they are opportunities that the proposed capability would open, not discoveries already in hand.

The concept would build on dedicated beam-combination optics, large-format SPAD arrays, and high-dispersion spectroscopy. The paper notes that analyzing ELTI's massive, multidimensional datasets will require significant computational power and sophisticated algorithms, with GPU-accelerated AI methods expected to enhance processing speed and scientific return. It also connects to ongoing Spanish investments in SPAD technology, including a megapixel detector under a CDTI Pre-Commercial Public Procurement call for astronomy, and to visible intensity interferometry at the Gran Telescopio Canarias and the La Palma Quantum Interferometer.

For now, the idea lives in a white paper aimed at the 2040s. But it suggests that the next leaps in astronomy may come not only from collecting more light, but from measuring when each photon arrives.