
1 Introduction
Little is known about particle nature of dark matter (DM) even though the existence of
DM has been firmly confirmed by the astrophysical observations. Strongly interacting
massive particles (SIMPs) [1] are an interesting framework for the thermal relic of sub-
GeV DM: the thermal relic DM is determined by the freeze-out of 3 →2 processes.
Self-interactions of DM are generically sizable to get the correct relic abundance in this
framework, and it leads to a large 2 →2 self-scattering. The small-scale structure of the
Universe may indicate the sizable self-scattering of DM [2] (see Ref. [3] a review).
Dark sector is a hypothetical sector where DM resides, and has its own gauge dynamics.
As a consequence of dark strong dynamics, the dark sector would consist of composite
particles (dark hadrons) at the low-energy scale as with the SM hadrons [4–51] (see
Ref. [52] for a review). Ref. [53] proposed a model of the SIMP framework where the dark
pion is identified as DM. The dark pions arise as the pseudo-Nambu-Goldstone boson
(pNGB) from the strong dynamics in the dark sector. The dark pions have the 2 →2
self-interaction and the 3 →2 number-changing process induced by the Wess-Zumino-
Witten (WZW) term [54, 55].
The chiral perturbation theory (χPT) describes the interactions among dark pions.
The pion self-coupling, which is defined by the ratio of the pion mass and decay constant
mπ/fπ, determines the size of the pionic scattering processes. The 2 →2 self-scattering
interaction arises at the leading order terms of O(m2
π/f2
π), while the 3 →2 number-
changing interaction via the WZW term appears as an O(m5
π/f5
π) term of χPT. The pion
self-coupling is found to be larger than unity in order to explain the relic abundance
and to evade constraints on the self-scattering cross section [53]. Meanwhile, we cannot
validate the perturbative expansion of χPT unless mπ/fπ.4π. We would not be able to
ignore contributions from resonances and the higher-order terms of the chiral Lagrangian
for the pion self-coupling near the na¨ıve perturbative bound. For the consistent treatment
of the chiral expansion, Ref. [56] has discussed the impact of the higher order of the chiral
Lagrangian on SIMP scenarios.
We may encounter the other bound on the self-scattering cross section of dark pions
in the SIMP models even when the perturbative expansion of χPT is valid. The unitarity
of the S-matrix imposes constraints on the partial-wave amplitude. The partial-wave
amplitude for the 2 →2 self-scattering is T'm2
π/(32πf2
π) at the tree-level, and the
perturbative unitarity places an upper bound 1/2βon ReTwith βcorresponding to
velocity of pions. The perturbative unitarity bound mπ/fπ.4√π/√βgets weaker at the
cosmic structures due to the DM velocity at maximum of ∼10−2, while this bound can be
important for the annihilation mechanism determining the current relic abundance since
dark pions are semi-relativistic. In other words, there are two bounds on the pion self-
couplings: one originates from the limitation of the perturbative expansion and another
is the perturbative unitarity of the scattering processes. In the chiral limit (mπ→0),
the perturbative unitarity violation will be cured by resumming multiple rescattering
processes, which is known as the “self-healing” mechanism [57]. In this paper, we will
propose the improvement of 2 →2 and 3 →2 partial-wave amplitudes in a similar way
to the “self-healing” mechanism in the non-chiral limit since we focus on the dark pion
DM.
1