HARMONI Cuts 155 Instrument Configurations While Protecting Core Science

A 39-Meter Telescope Needs More Than a Camera

On Cerro Armazones in Chile, ESO is building the Extremely Large Telescope, a 39-meter-wide eye on the sky. When it begins operations by 2030, it will have a collecting area several hundred times larger than any existing ground-based optical or near-infrared telescope. But raw light is not enough. Astronomers need to split that light into colors, pixel by pixel, across tiny patches of sky.

That is the job of HARMONI, the ELT's integral field spectrograph. Instead of taking one image, it measures a spectrum for every spatial element in its field of view. The result is a three-dimensional data cube: two dimensions of position, one of wavelength. HARMONI covers 0.75 to 2.4 micrometers, the near-infrared.

A new arXiv preprint describes HARMONI after a recent rescope. The question is not whether the instrument can do ambitious science. It is whether it can do that science with a design robust enough to build, budget, and deliver.

When a Rescope Makes an Instrument Sharper

HARMONI was already planned as the ELT's workhorse near-infrared spectrograph. Adaptive optics—real-time correction of atmospheric blur—would sharpen its view. But the project underwent a rigorous rescope co-led by the consortium and ESO. The paper reports the new architecture, performance forecasts, organization, and schedule.

The simplification is striking. Achievable instrument configurations fall from 181 to 26. Motorized cryogenic mechanisms drop from 26 to 14, including four that are set at integration and never moved during operations. The instrument still has three main blocks: a calibration and relay system that feeds light from the telescope, a cold spectrograph assembly that slices and disperses light, and a natural guide star sensing system that measures atmospheric distortion.

One addition is a Pointing Camera. Mounted beside the integral field unit, it images a wide field about 20 arcsec from the IFU centre during science exposures. Matching stars and compact galaxies to astrometric catalogues gives absolute position calibration to better than 30 milliarcseconds. That absolute calibration, combined with sub-spaxel relative registration from the natural-guide-star arms, allows data cubes taken months apart to be stacked into one deep observation and compared with data from JWST, ALMA, and other facilities.

HARMONI offers six spectral settings. Four medium-resolution modes at R≈7000—where R is a measure of spectral resolving power—cover the Iz, J, H, and K near-infrared windows individually. Two low-resolution modes at R≈2750–3000 cover broad chunks: Iz+J from 0.75 to 1.35 micrometers, and H+K from 1.45 to 2.40 micrometers. The blue limit of the first low-resolution mode was extended to 0.75 micrometers, partly to capture kilonova spectra, where the full P-Cygni profile of r-process elements falls in that range.

Three Adaptive-Optics Modes, Three Ways to Fight Blur

In Single Conjugate AO, HARMONI uses a natural guide star near the science target, working down to R=12. A pyramid wavefront sensor measures atmospheric distortion. The forecast is a Strehl ratio—a measure of how much light lands in the sharp core of an image—of about 70% in the K band under median conditions.

High Contrast AO adds coronagraphic masks and a finer 3.9-milliarcsecond spatial pixel scale. For guide stars brighter than mᵥ=9 and the best quartile of atmospheric conditions, the paper predicts it can suppress starlight enough to detect a point source with a peak intensity ratio of 5×10⁻⁵ at less than 100 milliarcseconds from the star at 1.65 micrometers, in a single 60-second exposure. That contrast corresponds to Jupiter-mass planets at 1–2 astronomical units (the Earth–Sun distance) around stars within 20 parsecs.

Multi-Conjugate AO uses MORFEO, the ELT's dedicated MCAO module. Six laser guide stars and up to three natural guide stars sensed by HARMONI help correct turbulence over a volume of atmosphere. Under the paper's natural-guide-star selection criteria, the forecast is wide sky coverage: more than 75% of the sky accessible at Strehl ratios above 50% in K band under good conditions at the South Galactic Pole, rising above 90% of accessible fields at 30% Strehl.

From Exoplanet Air to Galaxies at Cosmic Dawn

The science cases span a huge range. At high redshift, rest-frame optical emission lines from star-forming galaxies are stretched into the near-infrared. At z≈2, HARMONI's ~10-milliarcsecond angular resolution could resolve structures on 50–100 parsec scales, comparable to individual star-forming clumps, HII regions, and compact bulges. The same spectral diagnostics reach galaxies at z≈2–3 and into the epoch of reionization (z > 6).

For exoplanets, H and K band spectra could map molecules such as CO, water, and methane in directly imaged giant planets through cross-correlation techniques. For nearby galaxies, it could measure stellar velocity dispersions around supermassive black holes, expanding the census of dynamically confirmed black hole masses. For resolved stellar populations, it could separate individual upper asymptotic giant branch stars in the Virgo cluster. It could also follow kilonovae from neutron-star mergers and map Solar System bodies such as the ice giants Uranus and Neptune.

The performance forecasts are ambitious. For a 10σ detection per spectral pixel in a 10-kilosecond exposure, using HARMONI+MORFEO under median conditions and the 6-mas spaxel scale, the paper estimates HARMONI reaches AB≈23.5–24 per spectral pixel in R≈7000 mode, and AB≈25 in the H+K low-resolution mode at its most sensitive wavelengths. Compared with JWST/NIRSpec at comparable spectral resolution (around R≈2700–3000), the forecast is 1.5–2 magnitudes more sensitive for point sources—roughly a factor of 15–60 shorter exposure time—at about six times finer angular resolution.

The Forecast Is Not Yet the Sky

Those numbers are predictions, not observations. The completed HARMONI instrument has not been built or tested on the sky. Its performance will depend on atmospheric conditions, guide star availability, adaptive-optics stability, and the challenge of combining long integrations. The paper describes expected and simulated performance based on the current design.

The project has a new path forward. After an ESO Science Technical Committee meeting in April 2026, the rescoped concept was endorsed and then approved by the ESO Council at its 9–10 June 2026 meeting. A Rescope Design Review is scheduled for November 2026. The baseline schedule targets a Preliminary Acceptance Europe milestone in November 2035, with options aiming for 2034. The paper lists about €31.2 million in consortium hardware cost, excluding ESO-supplied detectors, and roughly 1,400 person-years of total FTE effort. The new principal investigator is Professor Jim Dunlop of the University of Edinburgh, who formally took up the role on 1 July 2025.

A Leaner Instrument for a Larger Sky

The rescope did not shrink HARMONI's ambition. It made the ambition more buildable. By cutting configurations and cryogenic mechanisms while protecting high-sky-coverage adaptive optics and full near-infrared spectral coverage at intermediate resolution, the team is betting that reliability matters as much as raw power.

If the forecasts hold, HARMONI will be a machine for turning tiny patches of sky into spectra, and spectra into stories about planets, black holes, and the first galaxies. The sky is waiting; the instrument must still be delivered.