Two hydrogen clouds near the Whirlpool galaxy, each holding roughly 3 million solar masses of gas and showing no stars, are the strongest candidates yet for a predicted class of starless dark-matter halos. The findings, published in Astronomy & Astrophysics on August 4, draw on data from China’s FAST radio telescope and introduce the concept of a reionization-limited H I cloud, or RELHIC — a small halo that retained gas after the cosmic dawn but never cooled enough to form stars.
But the clouds sit in a galaxy system known to be violently interacting. The same data cannot rule out that they are tidal debris torn from the Whirlpool’s disk rather than primordial halos. The distinction will require higher-resolution observations FAST cannot provide alone.
Each cloud holds about 3 million solar masses of hydrogen. They are compact, roughly spherical, and show narrow spectral lines around 20 kilometres per second. In deep optical images, they are invisible — no stars, no faint smudge of a galaxy, nothing. That is exactly the profile a reionization-limited H I cloud should have, if such an object exists outside a simulation. The catch is that they sit 70 to 90 kiloparsecs from the centre of the Whirlpool galaxy, a system famous for its ongoing gravitational dance with a companion. In that environment, gas can be pulled into long tidal streams that mimic the shape of a primordial halo at low resolution. The two clouds, designated Cloud N and Cloud S, are either the first real evidence of a population of dark-matter halos that the standard cosmological model predicts should be everywhere — or they are debris from a galaxy tearing itself apart. The telescope that found them cannot yet say which.
A pair of clouds that almost fit the model
The search was methodical. The FEASTS survey, using FAST‘s 21-cm observations, scanned 55 galaxies for compact hydrogen clouds lacking optical counterparts. Only two objects passed the filters. Cloud N and Cloud S each appear as unresolved blobs of neutral hydrogen at projected distances of 70–90 kiloparsecs from M51’s centre. The host dark-matter halos are estimated at 3.7 ± 0.4 billion solar masses, with gas fractions matching theoretical predictions for RELHICs.
The RELHIC framework, introduced by lead author Qingze Chen of the National Astronomical Observatories, Chinese Academy of Sciences, describes a dark-matter halo that never accreted enough gas or cooled sufficiently to ignite star formation. The ultraviolet background from the first galaxies kept the gas too hot to collapse. The signature is a compact, roughly spherical cloud with narrow line widths and no stars. The distinction between a primordial cloud and a tidal fragment is a matter of shape, motion, and environment — a checklist the visual below breaks down.
The ambiguity is the story. Chen and his collaborators stress that while the clouds are strong candidates, the tidal-debris alternative cannot be dismissed because M51 is interacting with its companion NGC 5195. The system’s tidal tails and bridges are well-documented. A higher-resolution radio observation — with the Very Large Array or the Square Kilometre Array — could resolve the clouds’ internal structure and reveal whether they are spherical or elongated. Chen points out that Cloud N may already show a velocity gradient inconsistent with a simple equilibrium halo, hinting at tidal stretching.
The RELHIC candidates are not the only FAST result this month. On September 1, 2026, a separate team published a study in Nature Astronomy that used FAST and the DESI instrument to stack 21-cm spectra from approximately 2.5 million galaxies. They found that the cosmic star formation rate has dropped by a factor of about 2.5 over the last 4.5 billion years, while the density of neutral atomic hydrogen declined by only a factor of 1.35. Peng Jiang, deputy director of NAOC, noted that the universe’s atomic reservoir did not disappear quickly enough to explain the fading stars.
The bottleneck appears to be the conversion of atomic hydrogen into dense molecular gas, not the exhaustion of fuel. That finding shifts the focus of galaxy-evolution models to the processes that regulate the atomic-to-molecular transition. For the M51 clouds, the question is similarly about a transition — the one that never happened. If they are RELHICs, they are fossil records of halos that failed to cross the threshold from atomic gas to stars. The data are suggestive. Are they definitive? Not yet.
The instrument that can’t see shape
FAST‘s single-dish beam at 21 cm is about 2.9 arcminutes across — wide enough to blur a compact cloud and a tidal filament into the same blob. The survey’s sensitivity is extraordinary, reaching column densities of a few times 1018 cm⁻², but spatial resolution is not its strength. The RELHIC classification hinges on morphology: spherical versus elongated. Without resolving the clouds, the evidence remains circumstantial.
The challenge of distinguishing dark-matter halos from interaction-born structures is not unique to M51. In the NGC 1052 group, astronomers have found a line of galaxies that appear to lack dark matter entirely, likely formed from a high-speed collision that separated dark matter from gas. That case, reported in June, shows that extreme interactions can produce objects that challenge the standard model just as forcefully as the missing RELHICs.
For Cloud N and Cloud S, the next step is clear: point an interferometric array at M51 and watch the clouds resolve. Until then, the Lambda Cold Dark Matter model’s prediction of a sky full of starless halos remains as elusive as the dark matter that shapes them.
Beyond the headline
The Bigger Picture
Taken together, the RELHIC candidates near M51 and the FAST–DESI neutral hydrogen measurements highlight a shift in galaxy-evolution thinking: the main challenge is no longer explaining how gas disappears, but how it fails to reach the dense, shielded molecular phase where stars form. Starless halos and gas-rich but low-efficiency galaxies occupy the low-mass frontier of structure formation, forcing ΛCDM models to track not just dark-matter assembly but the detailed thermodynamic and radiative histories of baryons across cosmic time.
The Science Gap
The key scientific gap exposed by these results lies between well-measured global trends and poorly understood small-scale physics. We now have precise curves for star formation and atomic hydrogen over billions of years, and plausible examples of starless halos, but lack direct constraints on how turbulence, magnetic fields, feedback, and environment regulate the atomic‑to‑molecular transition inside individual dark-matter wells. Bridging that gap will require combining 21‑cm surveys, molecular-line mapping, and high-resolution simulations tuned to objects like Cloud N, Cloud S, and Cloud‑9 rather than to bright, star-forming galaxies alone.
What Isn’t Being Said
Most coverage of these FAST findings emphasizes dark matter and star-formation decline but leaves out the methodological stakes. Reliance on stacking and single-dish beams in crowded environments means that classification of clouds as RELHICs or debris hinges on assumptions about unresolved structure and background subtraction that are rarely foregrounded. Acknowledging those limits changes the narrative: these are not just striking confirmations or refutations of theory, but stress tests of how far current radio-survey techniques can push into the dark and dim universe before interferometric follow-up becomes indispensable.
What this means for the people who study the dark universe
For astronomers mapping the invisible half of galaxy formation, the M51 clouds present a diagnostic challenge with immediate consequences.
- Western astrophysicist specializing in galaxy evolution
You need to assess whether the RELHIC framework holds up against the tidal-debris alternative. The published masses and line widths are consistent with the model, but the velocity gradient in Cloud N is a red flag. Consider proposing time on the VLA or MeerKAT to map the clouds’ kinematics at arcsecond resolution. If they are tidal, they will constrain the interaction history of M51; if they are primordial, they will be the first direct evidence of the starless halo population.
- Science policy analyst focused on global astronomy infrastructure
FAST’s rising output — two major papers in a single month — underscores China’s growing capacity in radio astronomy. The surveys are funded by the National Natural Science Foundation and the Ministry of Science and Technology, with explicit goals to lead in 21-cm cosmology. For Western funding agencies, this means competitive pressure: the next generation of deep H I surveys will require comparable investments in the Square Kilometre Array and its pathfinders to maintain scientific parity.
- Cosmology researcher utilizing 21-cm data
The FAST-DESI result that star formation efficiency declined while atomic gas persisted should be integrated into your simulations. The observed factor of 1.35 decline in Ω_HI over 4.5 Gyr, against a 2.5-fold drop in star formation rate, tightens constraints on feedback and gas accretion. Use these numbers to recalibrate the atomic-to-molecular transition in models like IllustrisTNG or EAGLE.
- Graduate student in observational astronomy
The methodology of stacking 21-cm spectra from 2.5 million galaxies, using DESI redshifts as a prior, is a template for your own work. The RELHIC search criteria — compactness, narrow line widths, no optical counterpart — are also something you can apply to archival data. The next step is to find more candidates in quiescent environments where tidal contamination is lower, perhaps in the FASHI or HD² survey fields.
Explainer
- Reionization-limited H I cloud (RELHIC)
- A theoretical class of small dark-matter halo that retained neutral hydrogen after the epoch of reionization but never formed stars. The ultraviolet background from the first galaxies kept the gas too hot to collapse, leaving a compact, roughly spherical cloud of atomic hydrogen detectable at 21-cm wavelengths. The two clouds near M51, if confirmed, would be the first observational examples.
- FAST
- The Five-hundred-meter Aperture Spherical Telescope, located in Guizhou, China, is the world’s largest single-dish radio telescope. Operated by the National Astronomical Observatories of the Chinese Academy of Sciences, it is sensitive to neutral hydrogen’s 21-cm emission and has conducted large-scale surveys like FEASTS and HD². Its single-dish resolution is limited, but its sensitivity allows detection of faint, extended H I clouds.
- Lambda Cold Dark Matter (ΛCDM)
- The standard cosmological framework, describing a universe dominated by dark energy and cold dark matter. It predicts that structure forms hierarchically, with small dark-matter halos merging to form galaxies, but not all halos become luminous — some remain starless. The RELHIC candidates test this prediction directly.
- DESI
- The Dark Energy Spectroscopic Instrument, mounted on the Mayall Telescope at Kitt Peak, Arizona, measures precise redshifts of millions of galaxies. The FAST-DESI collaboration used DESI’s redshift catalog to stack 21-cm spectra from about 2.5 million galaxies, enabling the measurement of cosmic neutral hydrogen density over time. Its data are critical for studying baryon cycles and dark energy.
- Whirlpool Galaxy (M51)
- M51 is a grand-design spiral galaxy located about 23 million light-years away in the constellation Canes Venatici. It is interacting with its smaller companion, NGC 5195, which has produced prominent tidal tails and triggered star formation. This interaction complicates the interpretation of any gas clouds found nearby.
- 21-cm line
- The 21-centimetre line is the hyperfine transition of neutral atomic hydrogen, emitted at a radio wavelength of 21 cm. It is the primary tool for mapping the distribution and kinematics of hydrogen gas in galaxies and intergalactic space. FAST’s surveys use this line to detect clouds like the M51 candidates.
RELATED STORIES
