A Proposed Spectrograph Could Let the ELT See What JWST Can Only Blur

At redshift 2.66 (a measure of cosmic distance and time), a galaxy called GLASS-180009 is already dead. It contains about 40 billion solar masses in stars, most of which formed in a brief early burst. JWST spectra show neutral gas falling into it at roughly 19 solar masses per year, yet it makes almost no new stars, less than 0.2 solar masses per year. The puzzle is not whether JWST can see it. It can. The puzzle is that JWST cannot see inside it clearly.

NIRSpec, JWST's near-infrared spectrograph, samples the sky with 0.1-arcsecond pixels. At that distance, each pixel covers about 800 parsecs, while the galaxy's effective radius is about 1,000 parsecs. The central and outer stellar populations blur together. A concept study called SHARP proposes to change that by pairing a near-infrared spectrograph with ESO's Extremely Large Telescope.

A Sharper, Wider Eye Than JWST

ESO's ELT will have a 39-meter primary mirror, and its multi-conjugate adaptive-optics system MORFEO is designed to correct atmospheric turbulence over a field 160 arcseconds across. At a wavelength of 2.2 micrometers, JWST's point-spread function—the blur of a point source—has a full width at half maximum of about 70 milliarcseconds. The ELT's diffraction-limited core would be about 12 milliarcseconds, roughly six times sharper. But sharpness alone is not enough. Astronomers also need that sharpness over a wide field, and they need to observe many objects at once.

SHARP is a concept for a near-infrared spectrograph covering 0.95 to 2.45 micrometers. It is not a camera. It would split light into spectra, revealing which atoms and ions are present, how stars and gas are moving, and how old or metal-rich a stellar population is. Its design is driven by what MORFEO can do: deliver a large, uniformly corrected field at the ELT.

The 200-Parsec Question

To understand how galaxies assembled and shut down, astronomers want to observe giant molecular clouds. These gas clouds, typically 150 to 250 parsecs across, are the main star-forming units, the main metal factories, and the cradles of globular clusters. Resolving them across cosmic time requires an angular resolution of about 30 to 35 milliarcseconds. NIRSpec reaches such physical scales only for galaxies closer than redshift 0.1.

At higher redshifts, key spectral lines shift into the near-infrared. Wavelengths beyond 0.9 micrometers are needed for galaxies at redshift above 0.8, and beyond 1.8 micrometers for those above 2.5. The HeII 1640 line, a leading marker for the first generation of stars, requires near-infrared spectroscopy at redshifts above 8 or 9.

Several planned ELT instruments cover parts of this territory but leave gaps. HARMONI has a small field of view and is effectively a single-object instrument. MICADO's long-slit spectroscopy is also single-object. MOSAIC uses 0.2-arcsecond fibers and stops at 1.8 micrometers. ANDES is a single-object spectrograph with extremely high spectral resolution. SHARP aims at the missing combination: high angular resolution, near-infrared coverage, and many targets at once.

Two Instruments in One

SHARP would contain two units. NEXUS, a multi-object spectrograph, would deploy up to 30 configurable slits over a field 1.2 by 1.2 arcminutes. Its pixels would sample 35 milliarcseconds. Four cameras would cover the full 0.95-to-2.45-micrometer range simultaneously, with selectable spectral resolutions of about 300, 2,000, and 6,000 for extended sources and about 17,000 for point sources. Its moving slit bars could be centered precisely on each target. An inversion prism could rotate the slit field, aligning each galaxy's major axis with the slit.

VESPER, a multi-integral-field unit, would target brighter sources. Twelve deployable field selectors would sample a 21-by-40-arcsecond area, with 31-milliarcsecond spaxels, or spectral pixels. Each spaxel produces its own spectrum, creating a three-dimensional map of a small patch of sky. VESPER would cover 1.2 to 2.4 micrometers with spectral resolutions of about 3,000 for extended sources and 4,000 for point sources. NEXUS is optimized for the faintest objects; VESPER is optimized for mapping brighter, extended ones.

What the Concept Predicts

The paper defines SHARP's sensitivity using two difficult measurements. The first is mapping stellar population gradients in passive galaxies at redshift around 3, down to stellar masses of about 10 billion solar masses. A 10-hour VESPER integration would reach a surface brightness of about 23 to 23.5 magnitudes per square arcsecond in the K band, around 2.2 micrometers, at one-sigma uncertainty. That is enough to measure how age, metallicity, and possibly the stellar initial mass function change from a galaxy's core to its outskirts.

The second is the search for Population III stars. For a hypothetical Population III stellar system of one million solar masses in a 200-parsec cloud at redshift 10, the predicted HeII 1640 line flux is about 4 times 10 to the minus 20 ergs per second per square centimeter. NEXUS could reach that flux in 10 hours, according to the concept study. These are forecasts from models and an exposure-time calculator, not detections. If the emission is significantly weaker, even SHARP might not see it, but a non-detection would still set upper limits on how much primordial star formation was possible.

One Dead Galaxy, Many Tests

GLASS-180009 serves as an illustrative case. NEXUS's 72-by-72-arcsecond field would encompass the galaxy and its surroundings, allowing many neighboring galaxies to be observed at once. VESPER's 21-by-40-arcsecond area would sample the galaxy itself and some of its neighbors, resolving physical scales of about 250 parsecs. Such data could help determine whether the inflowing gas comes from companion galaxies or from intergalactic filaments, why star formation remains suppressed, and how the galaxy's stellar populations vary across its body.

The example is meant to show versatility, not to report a result from SHARP. The instrument does not yet exist.

The Caveats: A Concept, Not a Telescope

SHARP is a concept study. It has not been selected or built. It depends on MORFEO, and MCAO (multi-conjugate adaptive optics) performance depends on atmospheric conditions and the geometry of natural guide stars. MORFEO works best with three bright, well-spaced guide stars. If the constellation is asymmetric, relies on fainter stars, or has only one or two stars, the correction degrades. The atmospheric dispersion corrector mentioned in the design is also not studied in detail in this paper.

The paper is a design overview posted to arXiv and part of a special issue on SHARP science. It reports no new observations. Its science cases and sensitivity estimates rest on models, simulations, and an exposure-time calculator, and the Population III detection case is especially dependent on assumptions about star formation efficiency and emission strength.

Still, the ambition is clear. If built, SHARP would not simply see deeper than JWST. It would see finer and wider at once, turning blurs like GLASS-180009 into maps of stellar populations, gas, and neighboring galaxies. That is the kind of view needed to ask why some of the universe's earliest massive galaxies died while gas was still falling in.