What this covers
Sean Carroll interviews astrophysicist Lina Necib on the nature and distribution of dark matter in the galaxy. Necib makes the case that dark matter is almost certainly real—not merely ordinary matter we haven't spotted—by marshaling evidence far beyond galaxy rotation curves: the power spectrum of the cosmic microwave background, dwarf galaxies, galaxy clusters, and roughly fourteen independent observables all point to a new kind of matter that modified gravity struggles to explain. She then turns to what has become the central practical puzzle: where dark matter actually is and how fast it moves. This matters because detecting dark matter in a laboratory depends entirely on knowing its local distribution and velocity in our galaxy, yet that distribution is neither smooth nor simple. Recent events in the Milky Way—major mergers captured in exquisite detail by the Gaia space telescope's measurements of a billion stars—have left the dark matter lumpy, dynamically complex, and still incompletely mapped.
Necib explores the landscape of dark matter candidates across roughly 40 to 72 orders of magnitude in mass, from fuzzy axion-like particles to primordial black holes, and explains why direct detection experiments, collider searches, and indirect detection via annihilation signals all probe different slices of parameter space. She details how recent Gaia data on the Gaia Sausage—a massive galactic collision from six to ten billion years ago—suggests the local dark matter may move slower than long-standing equilibrium models assumed, possibly weakening constraints that ruled out whole classes of theories. The conversation covers detection techniques from xenon tanks shielded deep underground to stellar stream gaps shaped by dark matter clumps, and addresses whether dark matter might be self-interacting, which would reshape galaxy centers in ways astronomers can observe. Throughout, Necib frames dark matter research as detective work: physicists know with precision what dark matter is not, and each null result functions as a clue rather than a dead end.
Necib argues that dark matter is almost certainly a non-Standard-Model substance whose existence is supported by many independent observations, and that precise mapping of stellar motions (especially via Gaia) reveals a lumpy, dynamically interesting dark matter distribution that directly affects our ability to detect it.
- Multiple independent observables (rotation curves, CMB power spectrum, ~14 different probes) point to dark matter beyond modified gravity
- Detection requires knowing dark matter's local distribution and velocity, which mergers like the Gaia Sausage and structures like Nyx change
- The dark matter sector is likely complex, possibly with self-interactions, rather than a single simple particle
Galaxy rotation curves, cosmic microwave background, and roughly fourteen independent observables across scales all independently point to dark matter as new matter species.
- Stars at the edges of spiral galaxies rotate at nearly the same speed as stars near the center, contradicting the expectation that rotational velocity should drop with distance if only visible stellar mass contributed gravity, which implies either gravity must be modified or unseen mass exists.
“stars on the edges are rotating pretty much at the same speed to the stars closer to the center, which is very bizarre”
- The case for dark matter does not rest on galaxy rotation alone; the cosmic microwave background power spectrum independently shows that about 84% of the matter budget is dark, and roughly 14 different observables across scales (galaxy clusters, dwarf galaxies, the Milky Way) all point to a new species of matter, which is far harder to explain with modified gravity than rotation curves alone.
“we know that 84% of the matter budget of the universe is part of this dark matter that is different from the state of matter, different from electrons, different from protons”
- About 300,000 years after the Big Bang, as the universe expanded and cooled, electrons bound with protons to form neutral hydrogen, making the universe transparent so that light could travel freely; this released light is now detectable as the cosmic microwave background.
“at about 300,000 years after the Big Bang, after the beginning of the universe, there was light”
Astrophysical observations of stellar streams and dwarf galaxies constrain dark matter by revealing its clumpy structure and ruling out theories like warm dark matter.
- Necib discovered Nyx, a clump of stars co-rotating with the Sun but moving toward the galactic center, which could be explained either by something hitting the Milky Way and causing waves or by a separate merger; if it is a merger, it may have brought in dark matter forming a co-rotating 'dark disk' that would substantially change direct detection estimates.
“Nyx is that... It's a clump of stars that are rotating with the Sun... they also kind of have this wave of movement towards the center of the galaxy, which is very bizarre”
- Stars pulled from disrupted satellites form streams in the sky, and gaps in such streams (like in stream GD-1) can be explained by a clump of dark matter passing through and pulling stars away; measuring the size and speed of the required clump constrains dark matter theories, because some theories (such as warm dark matter) cannot produce low-mass clumps.
“if you have a clump of dark matter that just goes right through your stream and pulls a lot of stars with it, then that means that's how you would get one of these gaps in your stream”
- Warm dark matter moves fast enough to destroy very small galaxies by streaming through and heating them up while leaving larger objects intact, which 'cuts the power spectrum' by forbidding small clumps; therefore observing small dark matter clumps (for example via stellar stream gaps) would rule out warm dark matter models.
“if you have a clump of dark matter that just goes right through your stream and pulls a lot of stars with it, then that means that's how you would get one of these gaps in your stream”
- Dwarf satellite galaxies of the Milky Way are better targets for indirect dark matter searches than the galactic center because they have far fewer baryons (stars and gas) relative to their dark matter—a high mass-to-light ratio—so any annihilation signal is less likely to be confused with conventional astrophysics like pulsars, and signals from many dwarfs can be stacked to gain statistical power.
“these smaller galaxies have a lot less Baryons or a lot less stars and gas than the center of the galaxy, which means that if you see signal from them, it's probably actually coming out of the dark matter”
- Measuring how much dark matter is in a dwarf galaxy is very difficult because the only measurement available is the line-of-sight (Doppler) velocity of a few bright stars, leaving the sideways motions unknown, and it requires assuming the system is in equilibrium, spherical, and unaffected by the Milky Way's gravity—assumptions reminiscent of the 'spherical cow in a vacuum' joke.
“all I see is what we call the line-of-sight velocity that basically is the... It's basically the Doppler Shift”
Null results from collider and direct detection experiments are informative, ruling out parameter space and driving the search for new theories and technologies.
- Despite hopes that the Large Hadron Collider would reveal new physics beyond the Standard Model, it has repeatedly confirmed the Standard Model, which is disappointing for those seeking the failure that would reveal what is missing.
“every time our experimentalist friends would show up new plots and they're like, "and The Standard Model is right again." and you think like, "Oh, no!"”
- The failure to detect new particles at the LHC and WIMPs at underground detectors—which probe overlapping mass ranges—is itself an informative observable that rules out theories, and rather than being wasted effort it functions as a call for creativity, demanding new theories that explain all null results and new experimental technologies to probe previously inaccessible masses.
“when you see nothing that is an observable because that means it's ruling out something. It's not really just that it was waste.”
- Direct detection experiments have ruled out huge swaths of WIMP parameter space without finding dark matter, and they are becoming so sensitive that they will soon hit the 'neutrino floor'—detecting solar neutrinos as an irreducible background—which is significant because these experiments have until now run essentially background-free.
“soon these experiments are going to be so good, that they are going to be detecting... They're hitting something that we call the Neutrino Floor and they're basically going to see neutrinos instead of the Dark Matter”
- Building a viable dark matter theory is hard because any candidate must satisfy a large number of existing constraints and not be ruled out by current experiments; we don't know what dark matter is, but we know in many specific ways what it is not, making the work more like detective work than free speculation.
“whatever theory you can come up with has to actually... Has to not be ruled out by existing experiments which actually were quite a lot”
- Physicists actively want the Standard Model to fail because they know it is incomplete—it is an effective field theory valid only up to a certain energy, lacks a theory of neutrino masses, and does not address quantum gravity—so a point of failure would reveal the hint of what new physics is needed; contrary to the stereotype, the field generates too many new ideas rather than too few.
“we know that it's what we call an effective field theory so it addresses a small... It is valid up to a certain energy but not beyond”
Understanding dark matter's lumpy distribution is essential because it couples particle physics and astrophysics, affecting indirect detection, direct detection, and lensing signatures.
- Indirect detection searches for the particles—electrons, positrons, neutrinos, gamma rays—produced when dark matter annihilates or decays, so for example an unexplained excess of gamma rays from the center of the galaxy (where dark matter density should be highest) could signal dark matter, though it could also be conventional astrophysics.
“dark matter is going to either annihilate or decay into particles that we already know and then we see those particles”
- Gravitational lensing techniques (weak lensing, micro-lensing) can detect black-hole dark matter because a massive compact object bends light from background sources, but smoothly distributed particle dark matter would not produce these lensing signatures, so the absence of expected lensing events sets bounds on the abundance of black holes at each mass range.
“if your dark matter is made out of particles, you have to adjust and have different ways of detecting it versus if it was black holes then you can use these methods of weak lensing”
- Because dark matter is distributed lumpily rather than smoothly, its structure couples particle physics and astrophysics across vastly different scales: lumpiness can produce gamma-ray excesses for indirect detection, while dark matter velocity affects direct detection because faster dark matter deposits more energy in a xenon detector, so understanding the galactic dark matter distribution is essential to interpreting all experiments.
“if the dark matter is faster or slower that would affect how many of them you would see in direct detection experiments in the tanks of Xenon underground, because the faster the dark matter the more energy it might deposit in the xenons”
Dark matter parameter space spans forty to seventy-two orders of magnitude in mass, requiring many complementary experiments each sensitive to only part of the space.
- The dark matter parameter space is enormous, spanning roughly 40 to 72 orders of magnitude in mass depending on how you count, from fuzzy dark matter at about 10^-22 electron volts up to massive black-hole candidates, which is why many different experiments are each sensitive to only part of the space.
“the dark matter parameter space is huge somewhere between 40 and 72 orders of magnitude, depends on how you count”
- Axions (or axion-like particles) are attractive dark matter candidates because they were originally motivated by theoretical needs about how quarks get mass and then happen to also be natural dark matter candidates; they are much lighter than WIMPs, with QCD axions around 10^-6 electron volts, illustrating that dark matter candidates span an enormous range of possible masses.
“they are motivated by theoretical needs and then they happen to be naturally very good candidates for dark matter as well”
Dark matter is fundamentally different ordinary matter and interacts weakly with the Standard Model; MACHOs and primordial black holes cannot constitute all dark matter.
- Dark matter is a fundamentally different kind of matter that interacts only weakly with the Standard Model, not merely dim ordinary objects; the MACHO (Massive Compact Halo Object) hypothesis that it could be dim stars has been ruled out across most mass scales by experiments.
“There are a lot of these experiments that actually ruled out different scales of MACHOs and that kind of tells us that it's really... It's not just some object that is just a star that is just way too dim”
- Primordial black holes remain a partially viable dark matter candidate, with small unexcluded windows (such as around 15 to 30 solar masses), but most of the parameter space is ruled out by gravitational lensing surveys, making it much more plausible that black holes form only a fraction of dark matter rather than all of it.
“I think 15 to 30 solar masses is still possible, something like that so it is not a completely ruled out theory.”
Dark matter likely forms a rich self-interacting sector rather than a single simple particle, and discovery requires satisfying numerous existing constraints simultaneously.
- Cosmic muons created when cosmic rays hit the atmosphere pass through us constantly and, while usually harmless, have been documented to flip bits in computer clusters (requiring error correction) and contribute to DNA mutations.
“there is actually a whole set of research about cosmic muons flipping bits in computer clusters and that they need to kind of correct for that”
- The FIRE (Feedback In Realistic Environment) simulations build galaxies resembling the Milky Way by starting with dark matter particles, adding gas and stellar physics, and evolving the system over 13 billion years; because each simulation 'particle' represents a large clump (e.g. one star particle is about 7000 solar masses), researchers can track stars and dark matter to find correlations between where stars came from and the local dark matter, informing what experiments should expect.
“you start the really, really early times you have your dark matter particles and you teach them gravity... you let that evolve from very early on and at the end of the day, 13 billion years later, you end up with a galaxy that looks very much like the Milky Way”
- The Galactic Center gamma-ray excess found by the Fermi telescope (1-3 GeV, strengthened with more data by 2014) initially looked like dark matter because its emission appeared isotropic, but the competing explanation of unresolved pulsars now appears more likely, with the current status being that the excess is probably mostly pulsars, pending radio-telescope confirmation.
“It's probably mostly pulsars, all those that are confirmed.”
- If the local dark matter density is lower than assumed, more dark matter theories remain viable because there is more wiggle room in the experimental constraints; moreover, a non-trivial velocity distribution such as from a dark disk changes the shape of exclusion limits rather than just scaling them, affecting high-mass dark matter constraints differently from low-mass ones.
“If dark matter is a lot less, then a lot of more of our theories are still possible.”
- If dark matter self-interacts—two dark matter particles scattering off each other—then the dense centers of galaxies become less dense because particles kick each other out, converting the steep 'cusp' density profile predicted by collisionless simulations into a flatter 'core'; observing cores rather than cusps would be an astrophysical probe that dark matter has self-interactions even though it does not interact with ordinary matter.
“if there is a high density spot and they're interacting too much, they will kind of kick each other out and that would drop the density of that part and this is what we call the core versus cusp”
- Given how complex the Standard Model is, it would be surprising if dark matter were just a single particle with a one-on-one interaction; Necib expects the dark sector to be a rich sector with interesting self-interactions, though she acknowledges this is an intuition without strong prior information rather than an observational result.
“I would expect that the dark matter sector and it's a whole sector, it's not just one particle for example, is going to be complex, it's going to have a very interesting interactions”
- Particle colliders can detect dark matter production by looking for missing energy: by conservation of energy and momentum, a stream of particles going one way with nothing balancing it on the other side signals an unseen particle, though neutrinos produce the same signature and must be eliminated, and the detection emerges statistically from accumulating many similar events rather than a single clear signal.
“you would see what we call missing energy in your detector”
- Direct detection experiments use multi-ton tanks of xenon placed deep underground (often in mine shafts under mountains) so that incoming dark matter rarely knocks an electron off a xenon atom, which is then pulled by an electric field to register a signal; the experiments must be deep underground to shield against cosmic-ray muons that would otherwise mimic dark matter signals.
“you're waiting for one Dark Matter particle, out of the many that might be going through your experiment, one of them to hopefully interact”
- Ground-based Cherenkov telescopes detect very high-energy gamma rays by catching the particle shower they create when interacting with the atmosphere, where the resulting particles move faster than the speed of light in the medium (such as water), even though nothing exceeds light speed in a vacuum.
“they end up moving in the medium of the experiment, usually water, faster than the speed of light in water. Nothing goes faster than the speed of light in a vacuum but in media, it actually can happen.”
Gaia space telescope measurements of stellar positions, distances, and motions enable dark matter research by revealing galactic structure and local dark matter properties.
- The Gaia space telescope, launched in December 2013, is revolutionizing dark matter astrophysics by measuring the positions, distances, and proper motions of one billion stars—about one percent of the Milky Way—with a subset of about seven million stars (in the 2018 second data release) having full 6D phase-space kinematics, vastly exceeding earlier missions like Hipparcos.
“Gaia is this space telescope that was launched in December 2013 with a goal of giving us the positional measurements so the motion and distances of one billion stars and this is one percent of the Milky way”
- Gaia's third data release, scheduled for the following year, will increase the number of stars with full 3D velocities from about seven million to roughly 100 to 150 million, dramatically expanding the dataset available for dark matter studies.
“for the third data release of Gaia, which is scheduled for next year... we're going to go from that seven million up to 100 to 150 million”
- Gaia's kinematic data revealed the 'Gaia Sausage,' a galaxy merger that smashed into the Milky Way six to ten billion years ago with a mass between 1% and 10% of the Milky Way, and analyzing the velocity of its stars suggests local dark matter may be moving slower than assumed, meaning experiments may have over-ruled-out parts of parameter space based on overly simple equilibrium models.
“there is such a merger that has happened somewhere between six and 10 billion years ago and smashed into our galaxy, brought in so many stars with it”
WIMPs remain popular dark matter candidates because their mass range naturally relates to weak force physics and offers the simplest hypothesis to test.
- Weakly Interacting Massive Particles (WIMPs) are a popular dark matter candidate in the mass range of roughly 1 to 100 GeV (between a proton and a gold atom), appealing because that range relates naturally to the weak force in the Standard Model and because a single simple particle with weak interactions is the easiest hypothesis to test.
“WIMPs would end up being in the mass of a one to a hundred GeV. That translates to somewhere between the mass of a proton to a mass of I think gold atoms”
- Standard neutrinos cannot constitute the dark matter because they are too 'hot'—moving too fast for their mass—so if they made up all the dark matter they would have streamed through and broken apart the cosmic structures we observe; only a heavier sterile (fourth) neutrino could potentially work.
“if they were all the dark matter, they would have destroyed a lot of the structure that we see is just because you can imagine that you have a clump of something and then the neutrinos are going so fast through it, they just break it apart”