How JWST puts the squeeze on light dark matter, for free

All throughout the Universe, there’s a massive puzzle whose solution remains unknown: the dark matter mystery. Within every large, high-mass system that we examine, including:

• spiral galaxies,
• elliptical galaxies,
• groups of galaxies,
• clusters of galaxies,
• cosmic filaments,
• and the large-scale cosmic web,

there simply isn’t enough normal matter to explain the gravitational signals we observe. From the internal motions of galaxies to the relative motions of galaxies within a cluster to the gravitational lensing signals generated by these objects to the clustering patterns of galaxies on the largest of cosmic scales, some novel type of mass that neither absorbs nor emits light — dark matter — must be present to consistently explain what we observe.

And yet, all of our efforts to directly detect dark matter have come up empty, with key signals from particle colliders, cosmic ray experiments, and possible signatures of dark matter annihilation all failing to appear. This leads to an interesting possibility: perhaps, if we can measure the signals from the Universe that would arise if dark matter had certain specific properties (e.g., mass, cross-section, interaction probability, decay rate, etc.), then we can either find those signals, giving us a clue as to what dark matter is, or use the lack of those signals to teach us what dark matter is not.

In a fascinating new study, Fermilab scientists Ryan Janish and Elena Pinetti have done just that, using data from the James Webb Space Telescope (JWST) to put the squeeze on light dark matter candidates. Dark matter is definitely out there, but certain scenarios involving axions and axion-like particles are now disfavored. Here’s how they did it.

The quantum fluctuations inherent to space, stretched across the Universe during cosmic inflation, gave rise to the density fluctuations imprinted in the cosmic microwave background, which in turn gave rise to the stars, galaxies, and other large-scale structures in the Universe today. This is the best picture we have of how the entire Universe behaves, where inflation precedes and sets up the Big Bang. Unfortunately, we can only access the information contained inside our cosmic horizon, which is all part of the same fraction of one region where inflation ended some 13.8 billion years ago.

When we take all the astrophysical data we have about the Universe, cumulatively, we can construct a consistent picture of everything we see just by applying the right rules and ingredients. That’s where the notion of our consensus cosmology comes from. Known as ΛCDM cosmology, its key ingredients include:

• Λ, the Greek Lambda, the symbol for Einstein’s cosmological constant, as a proxy for the dark energy in our Universe (68%, at present),
• CDM, for cold dark matter, the majority of the massive stuff in our Universe (27%, at present),
• plus all the “normal stuff” in our Universe, or everything made out of the known Standard Model particles, such as protons and neutrons, electrons, neutrinos, and photons (5%, at present),
• a hot Big Bang, occurring 13.8 billion years ago, representing the dense, nearly uniform state of our early Universe,
• and a period of cosmic inflation that preceded it, seeding our Universe with density imperfections at the start of the hot Big Bang.

The only other ingredient is simple: the fundamental laws of physics themselves, including the nuclear interactions, the electromagnetic interaction, and gravity as governed by Einstein’s general relativity. If you know the ingredients of the Universe, the initial conditions that it was born with, and the laws that govern it, then physics itself (and specifically astrophysics in this case) will predict what you ought to see when you look out at it later on, even all the way up to the present day.

This snippet from a structure-formation simulation, with the expansion of the Universe scaled out, represents billions of years of gravitational growth in a dark matter-rich Universe. Over time, overdense clumps of matter grow richer and more massive, growing into galaxies, groups, and clusters of galaxies, while the less dense regions than average preferentially give up their matter to the denser surrounding areas. The “void” regions between the bound structures continue to expand, but the structures themselves do not.

When it comes to dark matter, we have to rely on simulations to teach us what types of structures it will wind up forming. We’re fundamentally limited in the types of simulations we can run; even with powerful modern computers in 2025, with hundreds of billions or even trillions of particles in a single simulation, we can only simulate one specific set of scales at a time.

If we want to simulate the large-scale cosmic web, we can, but we don’t get very reliable results for smaller-scale structures (individual galaxies and below) from those simulations.

If we want to simulate an individual galaxy cluster, we can, and we can get pretty good results for the individual galaxies within it, but not necessarily for the substructure of those galaxies: the distribution of dark matter particles that clump up on scales smaller than a galaxy itself.

But if we go down to scales where we simulate a single individual galaxy, perhaps a galaxy comparable to the Milky Way in overall mass, we can not only get a grasp on what the structure of the galaxy ought to look like as far as how its mass (including dark matter) is distributed, but we can come to understand what type of dark matter substructure should exist within each such galaxy.

According to models and simulations, all galaxies should be embedded in dark matter haloes, whose densities peak at the galactic centers. On long enough timescales, of perhaps a billion years, a single dark matter particle from the outskirts of the halo will complete one orbit. Without a massive dark matter halo, galaxies would be smaller, lower in mass, and unable to hold onto the ejecta from stellar cataclysms.

Whereas you might have thought that dark matter would follow a smooth distribution within a galaxy, that’s only the coarse structure that comes out of our simulations. When we look at the finer and finer details that emerge, we start to see that there are these clumps that happen on smaller scales within any galaxy: hundreds of relatively large-sized clumps, thousands of intermediate-sized clumps, and hundreds of thousands or even millions of small-sized clumps.

You might dismiss that sort of work as being purely theoretical, as it is only derivable from simulations, but that would be a mistake. There are two lines of evidence that demonstrate that this dark matter substructure is very real. The first comes from looking at galaxy clusters, which don’t just have “spots of light” where the galaxies are, but also intracluster light: light from the stars that have been ejected or ripped out of their host galaxies, that trace out and follow the distribution of dark matter after enough time has passed. The distribution of this intracluster light clearly indicates the presence of this dark matter substructure, as the light distribution can’t be reproduced without it.

But perhaps even stronger evidence comes from systems of strong gravitational lenses with multiple (at least four) separate images of a background, serendipitously-aligned galaxy.

This image highlights six of the eight quadruply lensed systems first used to place the greatest model-independent constraints on dark matter’s temperature and mass from structure formation, alone, has also revealed the presence and distribution of dark matter substructure within the foreground lens system.

When we see these galaxies, or rather, these multiple images of the same galaxy, we have to recognize that many of their observed properties:

• brightness,
• position,
• and the amount of “stretch” they exhibit,

are heavily dependent on not merely the mass of the foreground galaxy acting as a gravitational lens by bending and distorting the space where they’re present, but how that mass is distributed, internally, throughout the foreground galaxy doing the lensing. These systems of strong lenses, known as quadruple lenses, have revealed the smallest dark matter “clumps,” or substructure, ever detected observationally.

So if that’s what we have around each galaxy — a coarsely smooth distribution of dark matter, in a halo, joined by smaller clumps of dark matter substructure — what does that imply for the nature of dark matter?

On its own, the sober answer is, unfortunately, “not all that much.” However, we can throw another piece of physics into play: the fact that, under many models of dark matter, it’s expected to either:

• self-interact, which produces photons,
• annihilate with itself, producing two photons whose energy equals the rest mass energy of the dark matter particle,
• or, in the case of an axion (or axion-like particles), to oscillate into photons directly.

This 4-panel graph shows constraints on solar axions, on the neutrino magnetic moment, and on two different “flavors” of dark matter candidates, all constrained by the latest XENONnT results. These are the best such constraints in physics history, and remarkably demonstrate just how good the XENON collaboration has gotten at what they do. Axions, like other dark matter candidates, have not yielded a positive direct detection signature yet.

It’s not mandatory that dark matter behave in one or more of these fashions, but it’s a scenario that’s not only allowed; in many models, it’s unavoidable. If dark matter is a particle that’s capable of these photon-producing interactions, that’s actually wonderful news. It would mean, first off, that dark matter isn’t completely dark, but can be detected — even if it’s indirectly, such as through annihilation into photons — with light, or electromagnetically.

Second off, if dark matter does self-annihilate, that means that by measuring the energy of the photons that dark matter annihilation produces, we can determine the mass of the underlying particle responsible for dark matter. After all, whenever a particle and its antiparticle, each of mass m, annihilate with each other, they produce two photons, each of energy E, as given by Einstein’s E = mc². If dark matter has the property of being its own antiparticle (a property that many bosons, including the photon and Z-boson, possess), then whenever two such particles interact, they’ll have a chance of annihilating and producing these photon signatures.

But it means something more as well: if we look for a photon signature that would arise if dark matter were its own antiparticle, and those photons aren’t there, that either tells us what dark matter isn’t or places constraints on what dark matter is and isn’t allowed to be.

Whether elementary or composite, all known particles can annihilate with their antiparticle counterparts. In some cases, particles are matter and antiparticles are antimatter; in other cases, particles and antiparticles are neither matter nor antimatter, and sometimes particles are their own antiparticle, which is anticipated for many candidates for dark matter. The typical result of this annihilation is the production of two, equal-energy photons that fly off in opposite directions to one another, where the energy of each photon is given by Einstein’s E =mc², where m is the rest mass of the annihilating particle.

In other words, we don’t necessarily need to build some sort of specialized, fancy detector or a dedicated experiment to look for certain forms of dark matter. If we understand the physics and astrophysics of the very real environments we have around and within galaxies in space, we can use the tools we already possess to search for dark matter. If it’s there, we’ll achieve the “holy grail” moment in cosmology and have some incredibly strong evidence for dark matter. And if it’s not there, we’ll still have advanced our understanding of the Universe in a profound way: by using the lack of electromagnetic signatures to constrain the remaining possibilities for what dark matter is allowed to be.

The most straightforward constraint comes from looking at the center of our own Milky Way: where the dark matter is densest locally, nearest to our own location. If dark matter were made of massive particles — particles somewhere around the mass of the Higgs boson — then when they annihilate, they would produce gamma-rays. We do, in fact, see gamma-rays from the galactic center, but they appear to be primarily due to pulsars within the Milky Way, not from dark matter. The absence, in fact, of a strong, high-energy emission line from the galactic center in gamma-rays gives us some very good constraints on the self-annihilation cross-section of WIMP dark matter: dark matter arising from a weakly-interacting, massive particle.

This map shows a 1-year view of the entire gamma-ray sky from NASA’s Fermi satellite. The growing-and-shrinking sources are active galaxies powered by supermassive black holes, but the transient “blips” that appear are the gamma-ray bursts that are so sought after, many of which are thought to also create black holes, albeit not the supermassive type. When the Moon enters the field-of-view of the telescope, it can temporarily become the brightest gamma-ray source in the entire sky, while the lack of gamma-rays from the galactic center constrain annihilating WIMP dark matter scenarios.

But down at much lower energies, not only below gamma-rays but below X-rays and even visible light, we come all the way down to infrared light. With less than a billionth of the energy-per-photon that gamma-rays possess, infrared light is rather unremarkable in many ways. However, there are two things it has going for it.

1. We actually have incredibly good infrared observatories, led by the NASA flagship JWST, which probes common, nearby structures such as galaxies to better resolution and fainter magnitudes than any observatory ever.
2. And even though WIMP-like dark matter is far too heavy and massive to produce infrared photons through annihilation via Einstein’s E = mc², scenarios involving light dark matter — such as axions or axion-like particles (or even light, sterile neutrinos) — would produce “extra” infrared photons if they were present.

In other words, just by looking at plain, normal, “flat” regions of sky, or regions of the sky where there are no known, detectable objects, we can perform a search for these decaying, annihilating, or oscillating dark matter particles of sufficiently low mass. Particularly by leveraging regions where spectroscopic data was taken with JWST, we can search for an infrared photon excess at any particular wavelength. If an excess is there, it’s a potential dark matter signature; if not, it tells you that dark matter isn’t.

This composite of 19 separate nearby, face-on spiral galaxies all located within 100 million light-years of ourselves is a result of the PHANGS program, where multi-wavelength observatories including Hubble, the VLT, ALMA, and now JWST are all taking data of the same objects in an effort to map out the stars, star-forming regions, gas, dust, and even diffuse atoms that exist within these objects. These galaxies are also all embedded in dark matter halos, so wherever “blank sky” can be found, JWST will be sensitive to any signals from decaying, annihilating, or oscillating dark matter within a particular mass range.

The great news is that JWST has imaged all sorts of different classes of object in space, including:

• nebulae within our own galaxy,
• nearby galaxies well beyond the Milky Way,
• and the extragalactic depths of deep space.

Many of these images include not just photometric data, but spectroscopic data as well: where the infrared light is broken up into its individual component wavelengths. Although these images were nominally taken to observe specific astronomical targets, most of the ones beyond our own galaxy contain components of the image that simply have “blank sky” within them: regions with no stars, gas or dust, galaxies, or other sources of photons within them.

These “blank sky” regions are normally used to calibrate the telescope: they simply measure the response of the telescope to a signal of “nothing,” where all you see is stray light and instrument noise. But if we look at these “blank sky” regions and add them all up, we could find an excess of emitted photons: photons that would preferentially exist at one specific wavelength if they revealed decaying or annihilating (or oscillating) dark matter particles! Even in regions that are far away from any known galaxy, we’re still embedded within the galactic halo of the Milky Way, so such photon-generating signals are still possible. With JWST’s range, we can be sensitive to either decaying, annihilating, or oscillation light dark matter in the mass range of 0.1 eV to about 4.1 eV, and the paper authors see no evidence for a positive signal.

The current constraints on the decay lifetime (top) and the axion-like coupling to photons (bottom) for light dark matter candidates. These results use only blank sky for data, and so will improve as long as JWST continues to observe anything at all.

This teaches us some profound lessons, even though JWST never gave any observing time to research dedicated to constraining dark matter in such a fashion.

• If dark matter is “light” in this fashion, and specifically in the narrow mass range of between 0.8 eV and 2.5 eV, then its lifetime must be at least ~10²⁶ seconds, or around a billion times greater than the age of the Universe.
• If dark matter is made of light axions or axion-like particles in this mass range, then its coupling to photons is extremely small: smaller even than constraints from stellar evolution allow.

From both total flux, which would be sensitive to a broad decay, and the flux at individual wavelengths, which would be sensitive to an annihilation or oscillation signal, we can place new, tightest-ever constraints on the existence, abundance, and stability of light dark matter.

The great thing about this work is the constraints we can place on dark matter with JWST scales with what we call integration time: the amount of time spent gazing at a particular (even an empty) region of sky. In the future, we might yet collect significantly greater amounts of data, particularly with a total observing lifetime for JWST that’s expected to exceed 20 years. The constraints on dark matter’s decay lifetime, in this relevant mass range, will increase as the square root of observing time, while the constraints on dark matter’s coupling-to-photons (or its self-interaction cross-section, in the case of annihilation) will increase as the fourth root (the square root of the square root) of observing time. For observations time that would be, for instance, 10,000 times as great as used in the present study, the decay lifetime can be better constrained by a factor of 100, while the coupling/cross-section will be better constrained by a factor of 10.

But most remarkable of all is this fact: just by leveraging existing JWST data, we can put the squeeze on what types of dark matter are still viable. For light dark matter candidates, these constraints will improve significantly over the coming years, with the only requirement being that JWST is still taking data from objects all across the Universe.