Editorial NexoScience
Before JWST launched, the record for the most distant galaxy with a spectroscopically confirmed redshift had been stuck for years around z ≈ 8, and every step beyond it took the better part of a decade. That frontier now sits at z = 14.32, and a team has just detected that same galaxy, JADES-GS-z14-0, in mid-infrared light with the telescope’s MIRI instrument. The complication is that the identical set of measurements supports two very different portraits of the object, and one of them does not fit any current model of how the first galaxies were built.
The detection required nearly a full day of staring at one patch of sky. The observations reached an on-source integration time of about 23.8 hours at an observed wavelength of 7.7 μm, among the deepest mid-infrared exposures ever taken. The galaxy is unusually bright for its distance, with a rest-frame ultraviolet absolute magnitude of roughly −20.81, and it sits only 0.4 arcsec from a foreground neighbour, so the two had to be separated exposure by exposure with detailed model fitting. The result was a flux density of 74.4 ± 5.6 nanojanskys at 7.7 μm against 46.9 ± 0.6 nanojanskys at 4.4 μm, an excess of 27.5 ± 5.6 nanojanskys that has to be explained.
That extra light most plausibly comes from oxygen and hydrogen glowing inside the galaxy. At z = 14.32, the 7.7 μm filter samples rest-frame wavelengths of about 4,400 to 5,700 Å, exactly where the Hβ line and the [O III]λλ4959, 5007 doublet fall. The modelling indicates that at least a third, and possibly all, of the excess comes from those nebular lines rather than from starlight. If so, this galaxy had already forged and spread metals through its gas roughly 280 million years after the Big Bang. The inferred line ratio [O III]/Hβ lands near 2.5, well below the typical value of about 6 measured in galaxies at z ≈ 8, which points to a gas that is enriched but still notably metal poor.
The conservative fit describes a young galaxy weighing about half a billion suns. Modelled with the code BAGPIPES, JADES-GS-z14-0 holds a stellar mass near 500 million solar masses, close to a tenth of the present-day Milky Way, packed inside a half-light radius of only 260 parsecs. Its mass-weighted stellar age comes out at roughly 15 million years, its star-formation rate at about 25 solar masses per year, and the combined equivalent width of [O III] and Hβ at some 370 Å. Compressed into that radius, the star formation rivals the most violent starbursts known in the nearby universe. Every one of those values is consistent with what cosmological simulations expect at the redshift frontier.
The second fit turns the object into something the universe should not have assembled yet. Run through the code Prospector, the same photometry prefers older stars with essentially no dust, and it attributes the 7.7 μm excess to a strong Balmer break from evolved stars instead of emission lines. That solution demands a stellar mass near 2,500 million solar masses, almost an order of magnitude above the BAGPIPES value, and only about three times below the maximum halo mass that abundance matching allows for galaxies at this epoch. Worse, it places the bulk of star formation at z ≈ 18 to 20, with mass-weighted ages of 80 to 100 million years and no recent star formation at all, which clashes with the bursty behaviour expected of early galaxies.
No cosmological simulation produces a galaxy like that at this moment in cosmic history. The FLARES simulations predict that galaxies above roughly one billion solar masses appear only below z = 14, and those above some three billion only below z = 13. The IllustrisTNG and THESAN projects push the limit further still, placing galaxies above one billion solar masses only below z = 12. The authors write that, were the Prospector properties correct, they would carry radical implications for models of galaxy evolution in the early universe, because no simulation predicts anything that massive at the redshift frontier, let alone at the z ≈ 18 to 20 where those stars would have formed.
The authors choose a side without closing the case. They consider the BAGPIPES solution more likely, since it agrees with theoretical predictions and since a top-heavy initial mass function, plausible in a universe this young and this metal poor, would both weaken the Balmer break and cut the inferred masses by a factor of three. They also trace the disagreement between the two codes to their different stellar libraries, evolutionary tracks and isochrones rather than to the data. But they state plainly that neither solution can be rejected on a formal basis. Settling it will take deep follow-up spectroscopy with MIRI’s low-resolution spectrometer, measuring Hβ, [O III] and Hα directly instead of inferring them from a single broadband filter.
Source: “Photometric detection at 7.7 μm of a galaxy beyond redshift 14 with JWST/MIRI”, by Helton et al., published in Nature Astronomy. DOI: 10.1038/s41550-025-02503-z.
