Sommerfeld Enhancement in Dark Matter Physics

• Sommerfeld enhancement is a non-perturbative phenomenon where long-range forces distort the wave function of slowly moving particles, modifying short-distance reaction rates.
• It is applied in dark matter research to quantify changes in annihilation and coannihilation rates through attractive or repulsive interactions and resonance effects.
• Methodologies include non-relativistic Schrödinger frameworks with analytic Yukawa potentials and numerical solutions to assess impacts on relic density and indirect detection.

Sommerfeld enhancement is the non-perturbative modification of a short-distance reaction rate by a long-range interaction acting on a non-relativistic two-body state. In dark-matter physics it usually denotes the change in annihilation or coannihilation rates caused by the distortion of the incoming wave function relative to a free plane wave, so that the probability density at short distance is enhanced or suppressed as the relative velocity decreases; related constructions also appear for final-state interactions and in above-gap absorption in two-dimensional Dirac materials (Qiu et al., 2024, Abe et al., 3 Mar 2026, Leppenen et al., 2021).

1. Definition, physical content, and scope

The central physical picture is that slowly moving particles spend enough time in a long-range potential for their scattering state to differ substantially from a plane wave. If the potential is attractive, the wave function can be enhanced near the origin and the annihilation probability rises; if the potential is repulsive, the wave function is depleted and the rate is suppressed. Several of the cited works stress that the effect is strongest at low relative velocity, which is why it enters both early-universe freeze-out and late-time indirect detection (Beneke et al., 2022, Ferrante et al., 16 Jul 2025).

In the most common formulation, the long-distance dynamics modify a hard annihilation process multiplicatively rather than replacing it. A representative expression is

$$\langle \sigma v\rangle = \sum_l S_l \cdot \langle \sigma v\rangle_{(l)}^{\rm pert}, \qquad \langle \sigma v\rangle_{(l)}^{\rm pert}\simeq a_l v^{2l},$$

where the partial-wave dependence of the perturbative rate is preserved and the Sommerfeld factor $S_l$ encodes the wave-function distortion (Jho et al., 31 Dec 2025). For asymmetric dark matter, the same structure is often written as

$$\langle \sigma v \rangle = a\,\langle S_s\rangle + b\,\langle v^2 S_p\rangle,$$

making explicit the separate $s$-wave and $p$-wave contributions (Qiu et al., 2024).

A common misconception is that the effect is always an enhancement and always belongs to the initial state. The literature supplied here shows both statements to be too narrow. Repulsive loop-induced quantum forces can suppress the rate rather than enhance it (Ferrante et al., 16 Jul 2025), and for annihilation into heavier unstable particles the same formalism can act in the final state, where the slow outgoing pair experiences its own long-range distortion near threshold (Abe et al., 3 Mar 2026).

2. Non-relativistic formulation

The standard treatment is a Schrödinger problem for the two-body relative coordinate. In electroweak dark matter, for example, the non-relativistic dynamics are written as

$$\left( - \frac{1}{2 \mu}\nabla^2 + V(r) - i \frac{1}{2}\Gamma(r) \right)\psi(r) = E\psi(r),$$

followed in the usual Sommerfeld treatment by neglect of the imaginary part and solution of the radial equation in partial waves (Abe et al., 15 Jun 2026). For a light force carrier the canonical potential is Yukawa,

$V(r) = - \frac{\alpha_X}{r} e^{-m_X r},$

or equivalently $V(r)=-\alpha e^{-m_\phi r}/r$ in the scalar-mediator notation (Beneke et al., 2022, Jho et al., 31 Dec 2025).

The partial-wave enhancement factor is extracted from the short-distance behavior of the distorted wave function. One explicit definition used for general orbital angular momentum is

$$S_l = \left| \frac{(2l+1)!!}{|\mathbf{p}|^l\,l!} \left. \frac{\partial^l R_{El}(r)}{\partial r^l} \right|_{r=0} \right|^2,$$

which reduces to the familiar wave-function-at-the-origin expression in the $s$-wave channel (Jho et al., 31 Dec 2025). In coupled-channel electroweak problems, the same logic is implemented through matrix Schrödinger equations for nearly degenerate neutral and charged states (Abe et al., 15 Jun 2026).

For massless or Coulombic force carriers, analytic expressions are available. In the asymmetric-dark-matter treatment with a massless mediator,

$S_l$0

so the enhancement grows strongly at low velocity (Abudurusuli et al., 2020). The thermal average is then performed over the non-relativistic velocity distribution, and in the massless force-carrier limit rational-function interpolations reproduce the exact thermal averages with less than $S_l$1 error for $S_l$2-wave and less than $S_l$3 error for $S_l$4-wave, allowing the standard analytic freeze-out method to remain accurate to within $S_l$5 (Iminniyaz et al., 2010).

3. Interaction structures and mediator sectors

The literature no longer treats Sommerfeld enhancement as a phenomenon tied only to a single tree-level Yukawa potential. The supplied works span single mediators, mediator towers, loop-induced quantum forces, neutrino-exchange forces, finite-density backgrounds, and finite-size composites. The interaction structure determines not only the size of the effect but also whether the rate is enhanced or suppressed, whether resonances occur, and which observables are most sensitive.

Setting	Interaction structure	Reported feature
Light vector or scalar mediator	Yukawa potential $S_l$6	Standard low-velocity enhancement (Qiu et al., 2024)
Narrow resonance plus long-range force	Product $S_l$7 at leading order	Complete factorization (Beneke et al., 2022)
Tower of mediators	$S_l$8	Off-resonant enhancement increased by about $S_l$9–$$\langle \sigma v \rangle = a\,\langle S_s\rangle + b\,\langle v^2 S_p\rangle,$$0 or $$\langle \sigma v \rangle = a\,\langle S_s\rangle + b\,\langle v^2 S_p\rangle,$$1 in the studied setup (McDonald, 2012)
Loop-induced quantum force	Vacuum potentials scaling as $$\langle \sigma v \rangle = a\,\langle S_s\rangle + b\,\langle v^2 S_p\rangle,$$2 or $$\langle \sigma v \rangle = a\,\langle S_s\rangle + b\,\langle v^2 S_p\rangle,$$3	Enhancement or suppression depending on operator structure (Ferrante et al., 16 Jul 2025)
Two-neutrino exchange	Naive $$\langle \sigma v \rangle = a\,\langle S_s\rangle + b\,\langle v^2 S_p\rangle,$$4 potential with UV-completion dependence	Strong sensitivity to short-distance completion (Coy et al., 2022)
Finite-density dark background	Density-dependent Yukawa-like effective force	Stronger enhancement in dense environments (Cheng et al., 30 Dec 2025)

For mediator towers, the potential becomes a sum of Yukawa contributions,

$$\langle \sigma v \rangle = a\,\langle S_s\rangle + b\,\langle v^2 S_p\rangle,$$5

and in the regime $$\langle \sigma v \rangle = a\,\langle S_s\rangle + b\,\langle v^2 S_p\rangle,$$6 the combined effect is an effective larger coupling. In the confining-CFT / warped-dual construction analyzed in the cited work, off-resonant enhancement is larger than the standard single-mediator result by about $$\langle \sigma v \rangle = a\,\langle S_s\rangle + b\,\langle v^2 S_p\rangle,$$7–$$\langle \sigma v \rangle = a\,\langle S_s\rangle + b\,\langle v^2 S_p\rangle,$$8 for $$\langle \sigma v \rangle = a\,\langle S_s\rangle + b\,\langle v^2 S_p\rangle,$$9 when the UV cutoff is near the Planck scale (McDonald, 2012).

Loop-induced quantum forces are more singular than Yukawa exchange. In the examples studied, two-scalar exchange yields an attractive potential behaving as $s$0 at short distance, attractive two-fermion exchange behaves as $s$1, and vector-vector two-fermion exchange is repulsive and behaves as $s$2 (Ferrante et al., 16 Jul 2025). The neutrino-force study reaches a parallel conclusion: the long-range two-neutrino-exchange potential scales as $s$3, but the physical effect on the Sommerfeld factor depends crucially on the UV completion because the short-distance behavior becomes $s$4 for a $s$5-channel completion or $s$6 for an $s$7-channel completion (Coy et al., 2022).

4. Resonances, saturation, and unitarization

The enhancement becomes especially large when the long-range potential supports a shallow bound state or a near-threshold resonance. In the Yukawa/Hulthén analyses used for halo phenomenology, the low-velocity behavior is commonly organized into perturbative, Sommerfeld, and saturation regimes. For the $s$8-wave example driven by a light CP-even scalar mediator, the intermediate regime gives $s$9 and hence $p$0, while below

$p$1

the enhancement stops growing and the cross section returns to the underlying $p$2-wave suppression; the cited paper gives a $p$3-wave resonance near

$p$4

(Jho et al., 31 Dec 2025)

When a narrow annihilation resonance is present in addition to the long-range force, the leading-order structure simplifies rather than becoming more entangled. The effective-field-theory analysis of resonant dark-matter annihilation shows that Sommerfeld enhancement and Breit-Wigner enhancement factorize completely at leading order in the non-relativistic and narrow-width expansions,

$p$5

with all non-factorizable long-distance effects canceling at leading power (Beneke et al., 2022).

Another recurring issue is the apparent violation of partial-wave unitarity near zero-energy resonances. In the Jost-function formulation, the conventional Sommerfeld factor is

$p$6

and if the long-range potential produces a zero of $p$7 close to threshold, the naive enhancement can become too singular. The renormalization-group treatment identifies the pathology as a secular failure of the perturbative expansion in the short-range annihilation sector and replaces the naive factor by a unitarity-consistent form with an improved Jost function, so that the pole acquires an imaginary part reflecting the decay width of the annihilating bound state (Watanabe, 13 Aug 2025). A subsequent comparison of unitarization prescriptions shows that cutoff-based, RG-improved, and Green’s-function methods coincide to leading order when the unitarity-preserving corrections are large, and that the regulated cross sections are independent of the UV regulator to a good approximation (Cimring et al., 6 May 2026).

The same resonance logic extends to unstable final states. For annihilation into heavier unstable particles, the width is incorporated through a complex energy argument in the final-state Schrödinger equation, and narrow bound states of the final-state pair generate resonant enhancements, while a large width washes them out. The formulation simultaneously includes off-shell final states below the nominal threshold (Abe et al., 3 Mar 2026).

5. Freeze-out, relic density, and coannihilation

In cosmology, Sommerfeld enhancement enters the Boltzmann equation through the thermally averaged annihilation cross section. For asymmetric dark matter with comoving abundances $p$8 and $p$9, the standard equation is

$$\left( - \frac{1}{2 \mu}\nabla^2 + V(r) - i \frac{1}{2}\Gamma(r) \right)\psi(r) = E\psi(r),$$0

with the conserved asymmetry written as $$\left( - \frac{1}{2 \mu}\nabla^2 + V(r) - i \frac{1}{2}\Gamma(r) \right)\psi(r) = E\psi(r),$$1 (Qiu et al., 2024). In this setting the enhanced annihilation rate reduces the relic abundance by enlarging the integrated annihilation term, and the depletion is especially visible in the antiparticle abundance $$\left( - \frac{1}{2 \mu}\nabla^2 + V(r) - i \frac{1}{2}\Gamma(r) \right)\psi(r) = E\psi(r),$$2 rather than in the majority species (Abudurusuli et al., 2020).

For asymmetric dark matter with a light but massive mediator, the Planck-normalized analysis yields lower bounds on the coupling and upper bounds on the mass. For the benchmark $$\left( - \frac{1}{2 \mu}\nabla^2 + V(r) - i \frac{1}{2}\Gamma(r) \right)\psi(r) = E\psi(r),$$3 GeV, $$\left( - \frac{1}{2 \mu}\nabla^2 + V(r) - i \frac{1}{2}\Gamma(r) \right)\psi(r) = E\psi(r),$$4, and $$\left( - \frac{1}{2 \mu}\nabla^2 + V(r) - i \frac{1}{2}\Gamma(r) \right)\psi(r) = E\psi(r),$$5, the smallest couplings in the near-symmetric limit are $$\left( - \frac{1}{2 \mu}\nabla^2 + V(r) - i \frac{1}{2}\Gamma(r) \right)\psi(r) = E\psi(r),$$6 for $$\left( - \frac{1}{2 \mu}\nabla^2 + V(r) - i \frac{1}{2}\Gamma(r) \right)\psi(r) = E\psi(r),$$7-wave and $$\left( - \frac{1}{2 \mu}\nabla^2 + V(r) - i \frac{1}{2}\Gamma(r) \right)\psi(r) = E\psi(r),$$8 for $$\left( - \frac{1}{2 \mu}\nabla^2 + V(r) - i \frac{1}{2}\Gamma(r) \right)\psi(r) = E\psi(r),$$9-wave, and for asymmetries $V(r) = - \frac{\alpha_X}{r} e^{-m_X r},$0 and $V(r) = - \frac{\alpha_X}{r} e^{-m_X r},$1 the reported maximal masses are about $V(r) = - \frac{\alpha_X}{r} e^{-m_X r},$2 GeV and $V(r) = - \frac{\alpha_X}{r} e^{-m_X r},$3 GeV in the $V(r) = - \frac{\alpha_X}{r} e^{-m_X r},$4-wave case and about $V(r) = - \frac{\alpha_X}{r} e^{-m_X r},$5 GeV and $V(r) = - \frac{\alpha_X}{r} e^{-m_X r},$6 GeV in the $V(r) = - \frac{\alpha_X}{r} e^{-m_X r},$7-wave case (Qiu et al., 2024). The kinetic-decoupling analysis further shows that plain $V(r) = - \frac{\alpha_X}{r} e^{-m_X r},$8-wave annihilation is essentially unaffected, whereas $V(r) = - \frac{\alpha_X}{r} e^{-m_X r},$9-wave and Sommerfeld-enhanced $V(r)=-\alpha e^{-m_\phi r}/r$0- and $V(r)=-\alpha e^{-m_\phi r}/r$1-wave annihilations remain sensitive after chemical freeze-out because the interaction rate continues to depend on velocity (Abudurusuli et al., 2020).

Sommerfeld enhancement also competes with modifications of the expansion history. In non-standard cosmology with an extra energy density $V(r)=-\alpha e^{-m_\phi r}/r$2, faster expansion leads to earlier decoupling and therefore raises the relic abundance, while Sommerfeld enhancement lowers it. The two effects oppose each other during freeze-out, and the cited study concludes that the allowed region in coupling and perturbative annihilation cross section is enlarged, even though the maximum allowed mass is essentially unchanged relative to the standard case (Rashidin et al., 30 Jan 2025).

Coannihilation and electroweak multiplet dark matter provide additional model-specific realizations. In neutralino–sfermion coannihilation, the Sommerfeld correction is of the order of several per cent in stau regions, while in stop regions it can reach a factor of $V(r)=-\alpha e^{-m_\phi r}/r$3 because of strong QCD attraction (Hryczuk, 2011). In the renormalizable spin-1 electroweak model studied in the supplied literature, inclusion of long-range electroweak Sommerfeld effects shifts the thermal relic window to

$V(r)=-\alpha e^{-m_\phi r}/r$4

within a perturbative regime, substantially above the familiar $V(r)=-\alpha e^{-m_\phi r}/r$5 TeV scale associated with scalar or fermionic electroweak triplets (Abe et al., 15 Jun 2026).

6. Extensions, phenomenology, and nonstandard realizations

Indirect-detection phenomenology often relies on the separation between freeze-out and present-day velocities. In the halo-excess model with a light CP-even scalar mediator, the three-regime structure of perturbative behavior, Sommerfeld scaling, and saturation permits a $V(r)=-\alpha e^{-m_\phi r}/r$6-wave annihilation model to remain thermal at freeze-out, grow to halo-scale rates, and then saturate in dwarf spheroidals. The benchmark

$V(r)=-\alpha e^{-m_\phi r}/r$7

gives

$V(r)=-\alpha e^{-m_\phi r}/r$8

with dwarf constraints avoided because the enhancement shuts off in the colder systems (Jho et al., 31 Dec 2025).

A different mechanism appears when the long-range force itself is medium induced. In the two-component model with a dominant fermionic dark matter particle $V(r)=-\alpha e^{-m_\phi r}/r$9 and an ultralight pseudoscalar background $$S_l = \left| \frac{(2l+1)!!}{|\mathbf{p}|^l\,l!} \left. \frac{\partial^l R_{El}(r)}{\partial r^l} \right|_{r=0} \right|^2,$$0, a finite-density loop effect produces an effective Yukawa-like potential whose coupling is proportional to the local background energy density,

$$S_l = \left| \frac{(2l+1)!!}{|\mathbf{p}|^l\,l!} \left. \frac{\partial^l R_{El}(r)}{\partial r^l} \right|_{r=0} \right|^2,$$1

The corresponding inner-Galaxy Sommerfeld factor can reach $$S_l = \left| \frac{(2l+1)!!}{|\mathbf{p}|^l\,l!} \left. \frac{\partial^l R_{El}(r)}{\partial r^l} \right|_{r=0} \right|^2,$$2, opening parameter space capable of fitting the Fermi-LAT Galactic Center GeV excess (Cheng et al., 30 Dec 2025). The broader quantum-force analysis reaches a related conclusion: coherent background corrections can generate temperature-induced resonances for massless mediators, and in a non-thermal bosonic background the effective annihilation rate can be either enhanced or suppressed because the background both creates a Coulomb-like force and shifts the effective dark-matter mass (Ferrante et al., 16 Jul 2025).

Finite-size dark matter changes the standard point-particle picture qualitatively. For puffy dark matter the potential is softened inside the overlap region, the relevant size parameter is $$S_l = \left| \frac{(2l+1)!!}{|\mathbf{p}|^l\,l!} \left. \frac{\partial^l R_{El}(r)}{\partial r^l} \right|_{r=0} \right|^2,$$3, resonance peaks shift, and increasing the radius suppresses the enhancement. In the model-independent analysis summarized in the supplied material, finite size turns a sharp zero-energy resonance into a finite-width band, while in the nugget-type example the band collapses back to a sharp resonance line once the internal structure is fixed (Xu et al., 29 Jun 2026). The self-interacting puffy-dark-matter study reports the same qualitative tendency and finds that for a large ratio between $$S_l = \left| \frac{(2l+1)!!}{|\mathbf{p}|^l\,l!} \left. \frac{\partial^l R_{El}(r)}{\partial r^l} \right|_{r=0} \right|^2,$$4 and $$S_l = \left| \frac{(2l+1)!!}{|\mathbf{p}|^l\,l!} \left. \frac{\partial^l R_{El}(r)}{\partial r^l} \right|_{r=0} \right|^2,$$5 the Sommerfeld factor approaches $$S_l = \left| \frac{(2l+1)!!}{|\mathbf{p}|^l\,l!} \left. \frac{\partial^l R_{El}(r)}{\partial r^l} \right|_{r=0} \right|^2,$$6 (Wang et al., 2023).

The concept is also not restricted to dark matter. In two-dimensional Dirac materials the absorbance above the gap is written as

$$S_l = \left| \frac{(2l+1)!!}{|\mathbf{p}|^l\,l!} \left. \frac{\partial^l R_{El}(r)}{\partial r^l} \right|_{r=0} \right|^2,$$7

where $$S_l = \left| \frac{(2l+1)!!}{|\mathbf{p}|^l\,l!} \left. \frac{\partial^l R_{El}(r)}{\partial r^l} \right|_{r=0} \right|^2,$$8 is the Sommerfeld factor produced by electron–hole attraction. The cited work finds strong enhancement near the interband edge, followed by rapid reduction at higher frequencies due to single-particle energy renormalization; in graphene, by contrast, the interaction correction is small and the Sommerfeld factor stays close to $$S_l = \left| \frac{(2l+1)!!}{|\mathbf{p}|^l\,l!} \left. \frac{\partial^l R_{El}(r)}{\partial r^l} \right|_{r=0} \right|^2,$$9 (Leppenen et al., 2021).

Taken together, these studies present Sommerfeld enhancement not as a single formula attached to a Yukawa potential, but as a general low-velocity framework for incorporating long-distance wave-function distortion into short-distance reaction rates. The supplied literature shows that its magnitude, sign, resonance pattern, and phenomenological meaning are controlled by mediator content, background occupation, channel structure, cosmological history, and, in composite scenarios, the spatial extent of the interacting objects (Beneke et al., 2022, Watanabe, 13 Aug 2025, Xu et al., 29 Jun 2026).