Self-Interacting Dark Matter (SIDM)
Dark matter that interacts with itself via some non-gravitational force, with cross sections tuned to affect galaxy interiors but not cosmological scales.
Placeholder for a 3D visualisation of Dark Matter Candidates. The interactive scene will land in Phase 3. Roughly 26% of the universe's energy density behaves like cold, non-baryonic, gravitationally clustering matter. Plain ΛCDM asserts that dark matter exists; this family asks what it is. Five candidates dominate the literature: WIMPs, axions, primordial black holes, sterile neutrinos, and self-interacting dark matter. Each has a different theoretical motivation, a different production mechanism in the early universe, and a different experimental signature today. None has produced a confirmed detection after decades of dedicated searches.
In one sentence
Dark matter that interacts with itself via some non-gravitational force, with cross sections tuned so the interactions thermalize the inner regions of dwarf galaxies (creating cores instead of cusps) but barely affect large-scale structure.
The claim
Standard cold dark matter is assumed collisionless: a particle that only interacts gravitationally. Self-Interacting Dark Matter (SIDM) relaxes this assumption. It posits some non-gravitational self-interaction with an elastic cross-section per unit mass σ/m in the range 0.1 to 10 cm²/g. At this scale, dark matter behaves like a fluid in dense halo cores but is virtually collisionless everywhere else, preserving ΛCDM's large-scale successes.
The original motivation, due to Spergel and Steinhardt in 2000, was the small-scale structure puzzle. Cold-dark-matter N-body simulations produce halo density profiles with sharp central cusps (Navarro-Frenk-White). Observed dwarf galaxies and low-surface-brightness galaxies have flatter cores. SIDM with σ/m ~ 1 cm²/g would thermalize the inner few kiloparsecs of small halos, converting cusps into cores naturally.
Whether SIDM is needed is contested. Baryonic physics (supernova feedback, AGN winds) can also produce cores in some galaxies, removing the need for a non-baryonic mechanism. Different galaxy classes give different answers, and cluster shapes constrain σ/m to be smaller than the value that affects dwarfs unless the cross-section is velocity-dependent. Active research focuses on particle implementations with light mediators that naturally produce the required velocity dependence.
The family stance
Dark matter is some form of non-baryonic, gravitationally clustering matter that is not in the Standard Model. Multiple specific candidate particles or objects are seriously researched. The case for 'something' beyond ordinary matter is overwhelming; the case for any one specific candidate is not.
Predictions
- Cored density profiles in dwarf and low-surface-brightness galaxies, with core sizes correlated with halo mass and velocity dispersion
- Reduced central densities and rounder inner halos in galaxy clusters compared to pure CDM, but a smaller effect than in dwarfs (because clusters have higher velocity dispersion and shorter halo crossing times relative to the SIDM mean free path)
- Possible offsets between dark matter and galaxies in merging cluster systems, depending on the velocity dependence of the cross-section
Evidence
- Several dwarf and LSB galaxy observations favor cored over cuspy profiles, consistent with σ/m ~ 1 cm²/g
- Simulations with SIDM reproduce dwarf-scale cores and have produced consistent predictions across multiple groups
- Particle models with light mediators (e.g., dark photons) naturally produce velocity-dependent self-interaction that can match both dwarfs and clusters with a single underlying coupling
Counterpoints
- Baryonic physics (star formation feedback, AGN) can also produce cores within CDM, reducing the necessity of SIDM
- Cluster shapes and ellipticity from gravitational lensing constrain σ/m below the value needed to affect dwarf cores, requiring velocity-dependent cross-sections that some models can produce but not all
- Missing-satellites and too-big-to-fail problems aren't fully addressed by SIDM alone; they require additional fixes
- Critics view SIDM as introducing a free parameter (the cross-section) rather than proposing a specific candidate particle
Variants in this family
▸Go deeperTechnical detail with proper terminology
Cross-section parameter: σ/m in the range 0.1 to 10 cm²/g, where σ is the elastic self-scattering cross-section. For a particle of mass m_DM = 1 GeV, this corresponds to σ ≈ 10^-25 to 10^-23 cm².
Thermalization scale: in a halo of velocity dispersion σ_v, the thermalization timescale is t_th ≈ (n_DM × σ × v_rel)^-1. For dwarf-galaxy velocity dispersion ~10 km/s, t_th < age of universe for σ/m ~ 1 cm²/g, so dwarfs reach the SIDM regime; for cluster velocity dispersion ~1000 km/s, t_th > age of universe, so clusters remain effectively collisionless.
Velocity dependence: light-mediator models give σ ∝ v^-4 at high velocities and σ ≈ constant at low velocities, with a transition velocity set by the mediator mass. This naturally produces large cross-sections in dwarfs (low v) and small cross-sections in clusters (high v).
Constraints: the Bullet Cluster and similar merger systems constrain σ/m < ~1 cm²/g averaged over cluster velocities. Halo-shape constraints from gravitational lensing give similar bounds. The most viable SIDM is therefore velocity-dependent.
References
Last reviewed May 17, 2026
Spotted an error? Have a source to add?
Prefer email?
You can also send a prefilled email with the variant URL already filled in.