Astronomers mining a quarter-century of publicly available NASA data found 84 objects that every major X-ray catalog in existence had missed — not because the data was inadequate, but because the standard analysis pipeline never looked below the energy threshold where these objects live. The discovery, published in Nature Astronomy on September 9, 2026, describes a previously unrecognized class of cosmic objects called hypersoft X-ray sources (HSSs) — and suggests they may simultaneously help resolve two of the most stubborn open problems in modern astrophysics: why there are not enough candidate progenitors for Type Ia supernovae, the explosions that undergird the evidence for dark energy, and why distant galaxies show signs of helium ionization that no known source population can fully account for.

“These clandestine X-ray sources are actually among the most energetic objects in galaxies, and they could be solving two cosmic mysteries at once,” said Mustafa Muhibullah of the University of Alabama, who led the study. The NASA Chandra press release confirms all three author quotes and the paper’s core findings.

What Standard Catalogs Have Been Missing for 25 Years

The Chandra X-ray Observatory, NASA’s flagship X-ray telescope, has operated since July 23, 1999 — accumulating a publicly accessible archive now spanning more than 25 years of observations across thousands of sources. Yet automated analysis pipelines, including the widely used Chandra Source Catalog version 2.1, apply a standard energy range of roughly 0.5 to 7 kiloelectronvolts (keV). Sources that emit most or all of their X-ray light below 0.5 keV — and especially below 0.3 keV — are effectively invisible to these catalogs. Not because Chandra cannot detect them, but because the default floor was never set low enough to capture them.

The team led by Muhibullah, co-authored by Jimmy A. Irwin (University of Alabama) and Rosanne Di Stefano (Center for Astrophysics | Harvard & Smithsonian), searched six well-observed nearby galaxies using archival Chandra data specifically in the 0.15–0.3 keV band. They required at least a 3-sigma detection in that soft band with no significant emission above 0.3 keV. The critical selection criterion was a photon ratio of more than 8:1 — more than eight soft-band photons detected for every one photon detected in the harder 0.3–1.0 keV band.

That ratio makes HSSs dramatically softer than the “supersoft X-ray sources” (SSSs) previously recognized as the softest class of X-ray binary — systems where a white dwarf steadily burns hydrogen accreted from a companion star, emitting X-rays with temperatures in the 15–100 electronvolt range. HSSs are cooler still, with spectral models constraining their blackbody temperatures to below approximately 21 electronvolts (roughly 230,000 Kelvin). At those temperatures, the bulk of their energy is emitted not in X-rays at all but in the extreme ultraviolet — a band of electromagnetic radiation almost completely absorbed by interstellar hydrogen and helium gas before it can reach any telescope. The only detectable signature an HSS leaves is the faint X-ray tail that bleeds just above the ultraviolet cutoff.

“By combing through the Chandra archive, we were able to eliminate what used to be a blind spot for telescopes,” said Di Stefano. “That’s how we found what appears to be a new class of cosmic objects with remarkable qualities.” Her quote appears in the official NASA Chandra announcement alongside confirmation of all the paper’s key findings.

Why the EUV Blind Spot Swallowed an Entire Object Class

The extreme ultraviolet — spanning roughly 10 to 120 electronvolts — represents one of astronomy’s most fundamental detection barriers. Interstellar neutral hydrogen absorbs EUV photons so efficiently that sources even a few hundred light-years away become nearly undetectable at most EUV wavelengths; sources in other galaxies, millions of light-years distant, are completely invisible at those energies. The only dedicated space-based EUV observatory ever flown was NASA’s Extreme Ultraviolet Explorer (EUVE), operational from 1992 to 2001, and its catalog contained just over 1,100 sources — all within or near the “local bubble,” the relatively low-density region of the Milky Way surrounding our solar system.

Objects peaking in the EUV in external galaxies could therefore only ever be detected by the faint X-ray tail they produce above the UV cutoff. Chandra is sensitive enough to detect that tail — but only if analysts look below 0.5 keV, and specifically below 0.3 keV where the HSSs concentrate their detectable signal. Every survey, every catalog, every automated pipeline that set its detection floor at 0.5 keV or higher missed every HSS in every galaxy it surveyed.

The team also had to contend with an instrument-specific complication. Over more than two decades of operation, Chandra’s Advanced CCD Imaging Spectrometer (ACIS) has accumulated a layer of carbonaceous contamination on its detector surfaces, which preferentially absorbs the lowest-energy X-ray photons and has progressively reduced the telescope’s soft-band sensitivity. To account for this, the researchers restricted their search to archival data collected before 2017 (through Cycle 18 of Chandra operations), when the soft-band sensitivity remained adequate. They calibrated the contamination correction using Chandra observations of the galaxy cluster Abell 1795 as a reference standard. These methodological details are described in the arXiv preprint of the main paper.

The galaxies surveyed — M31 (the Andromeda Galaxy, approximately 2.5 million light-years away), M101 (the Pinwheel Galaxy, about 21 million light-years away), and elliptical galaxies NGC 3115, NGC 3379, NGC 4697, and NGC 4472 — were chosen specifically for their long total exposure times in the archive, their proximity, and the variety of their morphological types. None of the 84 HSSs identified across these six galaxies appears in any existing catalog compiled from the same data.

Extraordinarily Luminous for Objects That Were Invisible

Despite emitting in a narrow sliver of the X-ray spectrum, the brightest HSSs are energetically extraordinary. The most luminous examples in NGC 4472 — a large elliptical galaxy in the Virgo Cluster, roughly 55 million light-years from Earth — already emit at luminosities of roughly 10^38 ergs per second in the 0.15–0.3 keV band alone. Because the great majority of their energy escapes as extreme ultraviolet radiation that cannot reach the detector, the true bolometric luminosity (total across all wavelengths) is far larger. Spectral modeling in the Muhibullah et al. preprint suggests the most luminous examples could be radiating at 10^39 ergs per second or more.

For comparison, our Sun radiates at approximately 3.8 × 10^33 ergs per second across all wavelengths. The brightest HSSs therefore outshine the Sun by more than 100 trillion times in total power output. They are comparable in luminosity to ultraluminous X-ray sources (ULXs), which are among the most energetic individual point sources known in external galaxies.

The difference is that ULXs are typically found in galaxies actively forming stars — because they are thought to be black holes or neutron stars accreting at extreme rates from massive young companion stars. HSSs, by contrast, appear prominently in the elliptical galaxies NGC 3115, NGC 3379, NGC 4697, and NGC 4472 — galaxies where active star formation has largely ceased. In gas-poor elliptical galaxies, where astronomers previously found few or no luminous X-ray point sources, HSSs are now the dominant luminous point sources — despite being completely undetected until now.

The sources also show a range of temporal behaviors. Some in NGC 4697 appear to have been detected persistently across an observation baseline of at least 11 years. At least one source in NGC 3379 appears to have switched off entirely between observations. This variability, documented in the main paper preprint, suggests that HSSs are not a single physically uniform class but a population of objects that share a spectral property — extreme softness — while arising from multiple different physical configurations.

What They Might Be Explaining: Two Longstanding Cosmic Mysteries

Type Ia supernovae — thermonuclear explosions of white dwarf stars — are the standard candles of modern cosmology. Their precisely calibrated luminosities allowed astronomers in 1998 to measure the distances to supernova-hosting galaxies precisely enough to determine that the universe’s expansion is accelerating — a finding that implied the existence of dark energy and was recognized with the 2011 Nobel Prize in Physics.

For this system of distance measurement to remain reliable, astronomers need to understand what objects produce Type Ia supernovae. The leading single-degenerate model proposes that a white dwarf in a binary system accretes mass from a companion until it approaches the Chandrasekhar limit of approximately 1.44 solar masses and ignites in a thermonuclear runaway. Supersoft X-ray sources — with their characteristic luminosities and temperatures consistent with a nuclear-burning white dwarf — were once the leading candidate progenitor class. But surveys found too few of them: SSSs fell short of the observed Type Ia rate by one to two orders of magnitude, as the Muhibullah et al. paper discusses in the context of the new discovery.

Adding HSSs to the tally changes that accounting. The discovery roughly doubles the estimated number of accreting white dwarf systems compared to previous surveys that searched only above 0.5 keV. Moreover, the 84 HSSs confirmed in six galaxies are almost certainly the tip of an iceberg — because the EUV absorption that makes them nearly invisible also means the detected sample represents only the fraction emitting enough X-ray tail to register, from galaxies close enough and with Chandra exposure times long enough to catch them.

“If we could find a way to spot these Type Ia supernova explosions before they go off, that would be really important,” said Irwin. “Right now, we study them after they’ve exploded, and astronomers have struggled to understand what is actually ignited.” His quote is confirmed in the official NASA Chandra announcement.

HeII emission lines — produced when photons with energies above 54 electronvolts strip electrons from helium atoms — have been observed in local and high-redshift galaxies for decades. These lines appear in spectroscopic observations from instruments including the James Webb Space Telescope and the Sloan Digital Sky Survey, in both star-forming galaxies and older stellar systems. Hot, massive stars alone cannot fully account for the HeII emission seen across the galaxy population — the photon budgets simply do not balance.

HSSs, with their estimated temperatures between 10 and 20 electronvolts and enormous bolometric luminosities, emit copiously at exactly the energies above 54 electronvolts needed to photoionize helium. They were simply not included in previous ionization models, because they had never been detected. The team’s analysis, available in the Chandra hypersoft sources paper, suggests that HSSs may constitute a previously overlooked but potentially major contribution to galactic ionization budgets — not just in nearby galaxies today but throughout cosmic history, when the star-forming and white dwarf populations were different from today’s.

What the Objects Actually Are (Still Unknown)

“We’ve never encountered a group of objects that act like this,” Muhibullah said. “Of course, the next step was to try to figure out what these things are.” His quote is confirmed in the NASA Chandra press release.

The lower-luminosity HSSs, particularly those in M31, have the clearest interpretation: several have been matched to known classical novae in optical catalogs, suggesting they are post-nova systems — white dwarfs that underwent a nova outburst and are still cooling, emitting from a contracting photosphere that passes through the hypersoft band on its way back to quiescence.

The higher-luminosity examples — those in massive elliptical galaxies radiating at the luminosities approaching 10^38 ergs per second in the narrow X-ray band alone — are harder to explain as white dwarfs, since no white dwarf has ever been observed radiating at such extreme output. Accreting black holes are an alternative, but typical black hole X-ray binary spectra are harder, not softer. In March 2026, researchers at the Sternberg Astronomical Institute in Moscow proposed a third possibility: disk accretion onto a binary black hole pair embedded in a hierarchical triple system. The Popov & Lipunova 2026 proposal describes a configuration in which two black holes of roughly 15 solar masses orbit each other at a separation of about 0.01 astronomical units (approximately 932,000 miles or 1.5 million kilometers), surrounded by a circumbinary accretion disk — a geometry that could naturally produce extremely soft, luminous emission consistent with what was observed.

One particularly striking feature is that HSSs are not preferentially found near star clusters. Standard X-ray binaries are almost universally associated with the regions where massive stars are forming or have recently died — because the compact objects that power them form from those massive stars. HSSs appear indifferent to stellar populations, showing up equally in old elliptical galaxies and young spiral arms. That spatial independence is itself a clue that the population is heterogeneous, representing more than one physical class of system united by their spectral softness.

The team has applied the label “hypersoft” as a phenomenological description — what these objects do spectrally — rather than a physical classification.

What Astronomers Will Look for Next

The most obvious next step is to search for more HSSs in other galaxies with archival Chandra coverage from before 2017. The same methodology — systematic search below 0.3 keV with the contamination correction — can be applied to any of the many nearby galaxies observed extensively by Chandra during its first 18 cycles of operation.

Ultraviolet follow-up using the Hubble Space Telescope is already planned. The team has a Hubble observing proposal targeting NGC 4697 and NGC 3379 with the Wide Field Camera 3 UV instrument (WFC3/UVIS F225W), seeking to detect the far-ultraviolet tails of HSSs as variable point sources against the galaxy background. At distances of tens of millions of light-years, the optical and ultraviolet counterparts of individual HSSs are expected to be far too faint for most current facilities — but Hubble’s sensitivity in the ultraviolet may be sufficient for the nearest examples.

The longer-term future of this research class lies with next-generation X-ray observatories with improved sensitivity at soft energies. ESA’s NewAthena X-ray observatory — endorsed by ESA’s Science Programme Committee in November 2023, planned for formal adoption in 2027 and launch in 2039 on an Ariane 6.4 rocket — will combine a large collecting-area mirror with improved instruments covering 0.2–12 keV. That expanded soft-band coverage, combined with far greater collecting area than Chandra, should allow far more comprehensive censuses of HSS populations across a much larger volume of the universe.

For now, the discovery stands as a demonstration that the most consequential findings in astronomy do not always require new hardware. A systematic, careful search of existing data — specifically, data that everyone had and nobody examined in quite this way — was all it took to reveal 84 objects that had been hiding in the archive for a quarter of a century.

Frequently Asked Questions

What makes hypersoft X-ray sources different from supersoft X-ray sources?

Supersoft X-ray sources (SSSs), known since the 1990s, emit X-ray light at temperatures between roughly 15 and 100 electronvolts — hot enough to produce a detectable spectrum across a broad band of soft X-rays. Hypersoft X-ray sources are cooler still, with temperatures below about 21 electronvolts. At those temperatures, the vast majority of their radiation escapes as extreme ultraviolet light, which is almost completely absorbed by interstellar gas before reaching any telescope. The only detectable signature is a faint X-ray tail below 0.3 keV — a part of the spectrum that standard catalog pipelines have never systematically analyzed.

Why can’t telescopes observe extreme ultraviolet light from distant galaxies?

Interstellar hydrogen and helium gas absorbs extreme ultraviolet radiation almost completely across distances greater than a few hundred light-years from Earth. The Lyman limit of neutral hydrogen — at a photon wavelength of 912 Angstroms, corresponding to about 13.6 electronvolts — creates a near-total absorption barrier for EUV photons at extragalactic distances. The only EUV observatory ever flown in space, NASA’s Extreme Ultraviolet Explorer (1992–2001), could only catalog about 1,100 sources, all nearby within our galaxy. Objects peaking in the EUV in other galaxies can only be detected through the faint X-ray tail they emit above the absorption cutoff.

Could hypersoft X-ray sources really change how we measure the expansion of the universe?

Not directly — but they could change a key input into those measurements. Type Ia supernovae are used as standard candles for measuring cosmic distances, which is how astronomers established that the universe’s expansion is accelerating and inferred the existence of dark energy. A long-standing problem has been that surveys found too few plausible progenitor systems — the accreting white dwarfs that could produce these explosions — to account for the observed rate. Adding hypersoft X-ray sources to the progenitor count could help resolve that discrepancy. If the progenitor population is larger than thought, it would confirm rather than undermine confidence in the Type Ia standard candle method — though the full accounting requires further research.

What could the Chandra finding tell us about what JWST is seeing in distant galaxies?

JWST has detected strong helium ionization signals in numerous galaxies across cosmic history that existing models cannot fully explain. Stripping electrons from helium atoms requires photons with energies above 54 electronvolts — squarely within the extreme ultraviolet range where hypersoft X-ray sources emit most of their energy. The Chandra team proposes that HSSs, previously invisible and therefore absent from all ionization models, could be a significant and previously unaccounted-for source of helium-ionizing photons throughout the universe’s history. If that is confirmed through future observations, the models astronomers use to interpret JWST’s spectroscopic data may need to be updated to include this newly recognized source population.