Ultra-Strongly Self-Interacting Dark Matter: From Phenomenology to Astrophysical Observables

M. Grant Roberts, Wolfgang Altmannshofer, Pierce Giffin, Stefano Profumo

First listed 2025-10-20 | Last updated 2025-10-20

Abstract

We develop a minimal, testable framework for two-component self-interacting dark matter (SIDM) in which a dominant, moderately self-interacting species coexists with an ultra-strongly self-interacting subcomponent (uSIDM). A light vector mediator induces velocity-dependent self-scattering, while early-universe dynamics - standard $2 \to 2$ annihilation supplemented by interconversion $χ_1χ_1 \to χ_2χ_2$ - determine the relic abundance analytically. From observations of dwarf and low surface brightness galaxy rotation curves, as well as strong cluster lensing, we place constraints on the microphysics parameters. From these constrained regions, we map the microphysics to effective \texttt{ETHOS} parameters and evolve the linear power spectrum in \texttt{CLASS}. We then confront the model with direct-detection constraints and place an upper bound on our parameter space. We identify a region where: (1) the SIDM dominant component attains $σ_{\rm{eff}}/m = 20 - 40~\text{cm}^{2}\text{g}^{-1}$ at dwarf velocities while satisfying cluster upper bounds $σ_{\rm{eff}}/m < 0.13~\rm{cm}^{2}\rm{g}^{-1}$; (2) a subpercent uSIDM fraction drives accelerated gravothermal collapse in early halos, providing seeds relevant to high-redshift quasar formation and ``little red dots''; and (3) the small-scale cutoff in the matter power spectrum remains consistent with Lyman-$α$ and satellite counts, but exhibits non-standard features, potentially discernible with future observations. The allowed space can be organized by the mediator-to-DM mass ratio and the late-time uSIDM fraction, with a narrow window singled out by the combined cosmological and astrophysical requirements.

Short digest

Introduces a minimal two-component SIDM model with a light vector mediator, analytic relic accounting (annihilation plus interconversion), and an ETHOS mapping evolved with CLASS, constrained by dwarf/LSB rotation curves and strong cluster lensing. The viable window features a dominant SIDM with σ_eff/m ≈ 20–40 cm^2 g^-1 at dwarf velocities while satisfying cluster bounds <0.13 cm^2 g^-1, plus a subpercent uSIDM fraction that undergoes rapid gravothermal collapse to seed early SMBHs and the “little red dots.” The small-scale matter power spectrum remains consistent with Lyman-α and satellite counts but exhibits distinctive, non-standard cutoffs that future data could test. The allowed space is organized chiefly by the mediator-to-DM mass ratio and the late-time uSIDM fraction, and is further bounded by direct-detection limits.

Key figures to inspect

• Figure 1 (process diagrams): Contrast number-conserving self-scattering, annihilation, and χ1χ1 ↔ χ2χ2 interconversion to see why conversion is rate-suppressed, motivating the simplified Boltzmann treatment for relic fractions.
• Figure 2 (cross-section hierarchy vs x): Verify that conversion is many orders below annihilation and that the non-relativistic expansion matches the full result at large x, underpinning the analytic yields and the freeze-out uSIDM fraction.
• Constraints panel (rotation curves + cluster lensing): Inspect σ/m as a function of velocity—look for the band giving 20–40 cm^2 g^-1 at v_dwarf and <0.13 cm^2 g^-1 at v_cluster—to see the microphysics region that survives data.
• Power spectrum figure (ETHOS → CLASS): Examine P(k) to locate the small-scale cutoff and any non-monotonic features compatible with Lyman-α and satellite counts; note scales where deviations from CDM are largest.
• Parameter-space map (m_med/m_DM vs f_u): Identify the narrow allowed window satisfying astrophysical and cosmological bounds and mark slices testable by sub-GeV direct detection under kinetic mixing.