---
title: Fermionic Asymmetric Dark Matter
url: https://www.emergentmind.com/topics/fermionic-asymmetric-dark-matter-adm
type: topic
---

# Fermionic Asymmetric Dark Matter

Searching arXiv for the supplied and closely related fermionic ADM papers to ground the article and citations.
Fermionic asymmetric dark matter (ADM) denotes dark matter whose present abundance is set by a particle–antiparticle asymmetry, analogous to the baryon asymmetry, with the dark state itself being a fermion. In the standard ADM logic, the abundance relation \(\Omega_{\rm DM}/\Omega_b = m_{\rm DM} n_X/(m_p n_B)\) implies a mass of order a few GeV if visible and dark asymmetries are comparable, but the fermionic ADM literature also contains composite baryonic realizations, heavier models in which the asymmetry densities are not comparable, and keV asymmetric warm dark matter in which the asymmetry mainly changes the phase-space distribution rather than fixing a GeV-scale mass [1203.5803, 1505.07410, 2405.10303].

## 1. Relic asymmetry and fermionic quantum numbers

In ADM, the relic abundance is not primarily set by thermal freeze-out of a symmetric particle–antiparticle population, but by a primordial asymmetry in the dark sector. For fermions, that asymmetry is naturally defined when the dark matter is non-self-conjugate. A recurring statement across the literature is that a Majorana fermion cannot carry a conserved particle–antiparticle asymmetry in the usual sense, whereas a Dirac fermion, or another complex fermionic state carrying a conserved dark quantum number, can do so [1203.5803]. This is why many fermionic ADM constructions are explicitly Dirac, or effectively Dirac before late-time symmetry breaking.

The standard abundance estimate is model-independent. If the mechanism linking the dark and baryon asymmetries conserves a combination of \(B-L\) and a dark charge \(X\), then \(m_{\rm DM}\sim (5-7)/Q_{\rm DM}\ {\rm GeV}\) is the characteristic scale, and in one explicit Dirac-type leptogenesis framework the fully asymmetric limit gives \(m_\chi \simeq 1.8~\mathrm{GeV}\) through \(n_\chi = \frac{79}{28} n_B\) and \(\Omega_{\rm DM}/\Omega_B \simeq 5\) [1203.5803, 2510.13723]. This does not fix all fermionic ADM models to the GeV scale. A plausible implication is that the GeV prediction is a common benchmark rather than a universal theorem, because heavier ADM can arise if the dark asymmetry is diluted or the dark number density is reduced relative to the baryon asymmetry, and keV asymmetric warm dark matter can use the asymmetry mainly to realize a degenerate Fermi gas [1401.7664, 2405.10303].

The distinction from bosonic ADM is structural. In fermionic ADM, compact-star support is governed by Fermi degeneracy pressure rather than Bose condensation, and this difference propagates into neutron-star accumulation, collapse criteria, and compact-star equilibria [1305.6908, 1809.08254]. One notable variant even starts with asymmetric Dirac dark matter and converts it into a Majorana fermion only after annihilations have frozen out, so the relic abundance is fixed while the state is Dirac, but the present-day particle avoids the dangerous vector \(Z\)-exchange characteristic of a Dirac electroweak fermion [1205.2844].

## 2. Microscopic realizations and mass generation

The fermionic ADM literature spans elementary singlets, mirror-baryon-like fluids, composite twin baryons, and sterile-neutrino-like warm dark matter. The following examples illustrate the range of constructions.

| Framework | Fermionic candidate | Characteristic feature |
|---|---|---|
| Minimal light fermionic ADM [1306.5878] | singlet Dirac fermion \(\chi\) | light real scalar mediator mixed with the Higgs |
| Fraternal Twin Higgs ADM [1505.07410] | spin-\(3/2\) twin baryon \(\Delta' \sim b'b'b'\) | \(m_{\Delta'} \approx 5\Lambda'_{\rm QCD}\) |
| Dirac-type leptogenesis with exact \(U(1)_L\) [2510.13723] | Dirac fermion \(\chi\) | \(m_\chi = \delta \Sigma v_S\), with \(m_\chi \simeq 1.8~\mathrm{GeV}\) in the fully asymmetric limit |
| Asymmetric warm dark matter [2405.10303] | RHN-like Dirac fermion \(N\) | \(|\mu| \gg T_{\rm dark}\) gives a degenerate Fermi gas |

In the minimal singlet construction, the Lagrangian is
\[
\mathcal{L} = i\,\overline{\chi}(\slashed{\partial}-m_{\chi})\chi +\frac{1}{2}\left(\partial_\mu \phi^\prime \partial^\mu \phi^\prime -m_{\phi^\prime}^2 {\phi^\prime}^2\right) -\kappa \overline{\chi}\chi \phi^\prime -V(H^\prime,\phi^\prime),
\]
with a light real singlet scalar \(\phi'\) mixed with the Higgs, so that the physical scalar eigenstates are
\[
h = (\cos\alpha)\,h' - (\sin\alpha)\,\phi',\qquad \phi = (\sin\alpha)\,h' + (\cos\alpha)\,\phi' .
\]
This is a minimal renormalizable realization of GeV-scale fermionic ADM in which the additional light scalar is introduced because a singlet fermion has no renormalizable coupling to Standard Model fields and the dimension-5 Higgs portal \(\mathcal{L}_5=\frac{\lambda}{\Lambda}|H|^2(\bar\chi\chi+{\rm h.c.})\) is too weak to remove the symmetric component if \(\Lambda \gtrsim {\cal O}({\rm TeV})\) [1306.5878].

Composite fermionic ADM appears naturally in the Fraternal Twin Higgs. There the central candidate is the twin baryon \(\Delta' \sim b'b'b'\), a spin-\(3/2\) fermion whose mass is dynamically set by confinement, \(m_{\Delta'} \approx 5\Lambda'_{\rm QCD}\), in a sector where \(\Lambda'_{\rm QCD} \sim 0.5 - 20\ {\rm GeV}\) is already restricted by Twin Higgs naturalness [1505.07410]. This is technically important because the ADM mass is not inserted by hand; it is set by a QCD-like scale.

A distinct class ties fermionic ADM to neutrino mass generation and baryogenesis. In the unified Dirac-neutrino model, an exact global lepton number \(U(1)_L\), a gauged \(U(1)_X\), and an unbroken \(\mathbb{Z}_4^D\) stabilize a Dirac fermion \(\chi\), with
\[
m_\chi = \delta \Sigma v_S,\qquad m_\nu \sim \epsilon \delta v_S,\qquad \frac{m_\nu}{m_\chi}\sim \frac{v_{\rm EW}}{v_S}.
\]
Heavy Dirac neutrino decays then generate equal and opposite asymmetries in the visible and dark sectors, with \(\epsilon_{\nu^h_1 \to \ell H} = - \epsilon_{\nu^h_1 \to \chi \phi^*}\) [2510.13723]. This makes the fermionic ADM asymmetry a literal lepton asymmetry stored in the dark sector.

At much lower mass, asymmetric warm dark matter uses a RHN-like Dirac fermion \(N\) with distribution
\[
f_N[E,\mu]= \frac{1}{\exp\left(\frac{E-\mu}{T_{\rm dark}}\right) + 1},\qquad 
f_{\bar{N}}[E,\mu]=f_N[E,-\mu].
\]
For \(\mu<0\) and \(|\mu|\gg T_{\rm dark}\), the anti-DM population dominates, \(n_{\bar N}\simeq g_N |\mu|^3/(6\pi^2)\), and the system becomes a degenerate Fermi gas [2405.10303]. This suggests that the fermionic ADM label covers both abundance-based and phase-space-based uses of asymmetry.

## 3. Symmetric-component depletion and hidden-sector interactions

A defining technical requirement in fermionic ADM is efficient removal of the thermally produced symmetric component. In the minimal singlet-scalar model, the dominant process is
\[
\chi\bar\chi \to \phi\phi,
\]
with
\[
\sigma v = \frac{\kappa^4 \cos^4\alpha \,(s-4m_\chi^2)}{24 \pi (2m_\chi^2-m_\phi^2)^4} \left(9m_\chi^4-8m_\chi^2 m_\phi^2+2m_\phi^4\right) \sqrt{1-\frac{m_\phi^2}{m_\chi^2}},
\]
and thermal average
\[
\langle \sigma v \rangle_x \simeq 3 \times 10^3 ~\text{pb} \left(\frac {10 ~\text{GeV}}{m_\chi}\right)^2 \left(\frac{30}{x}\right) \kappa^4 \cos^4\alpha .
\]
Imposing \(\langle \sigma v \rangle_{x_F} > 1~{\rm pb}\) gives \(\kappa \ge 0.2\) for \(m_\chi=17.1~\mathrm{GeV}\) and \(\kappa \ge 0.1\) for \(m_\chi=3.42~\mathrm{GeV}\) [1306.5878]. These lower bounds are not optional model details; they are the condition that keeps the relic abundance genuinely asymmetric.

A broader model-independent analysis emphasizes the same point. Successful ADM models need a light dark matter candidate, typically near a few GeV, and large annihilation cross sections to eliminate the thermally produced symmetric population. One common way to realize this is a lighter mediator with \(m_{\rm med} \sim 10\text{--}100~{\rm MeV}\), into which the dark matter annihilates efficiently; if that mediator decays into lighter dark-sector states rather than Standard Model particles, the hidden sector contributes dark radiation and potentially DM–DR scattering signatures in \(\Delta N_{\rm eff}\) and the matter power spectrum [1203.5803]. The same framework gives
\[
\left.\Delta N_{\rm eff}\right|_{\rm CMB} = \frac{13.56}{g_*^s(T_d)^{4/3}} \frac{(g_\ell+g_h)^{4/3}}{g_\ell^{1/3}},
\]
and parameterizes DM–DR momentum transfer as either
\[
\langle \sigma_{\rm DM-DR}v\rangle = Q_0\,m_{\rm DM},
\qquad 
\langle \sigma_{\rm DM-DR}v\rangle = \frac{Q_2}{a^2}\,m_{\rm DM}.
\]

Specific fermionic models realize this hidden-sector annihilation in different ways. In the Dirac-type leptogenesis model, the dominant annihilation channel is
\[
\chi\bar{\chi}\to \eta_I\eta_I,
\]
mediated by \(t/u\)-channel \(\psi_a\) exchange, and for \(m_\chi=1.8\,\mathrm{GeV}\), \(m_\psi=10\,\mathrm{GeV}\), and \(5\,\mathrm{MeV}<m_{\eta_I}<m_\chi/2\), a fully asymmetric regime is achieved once \(|f|\gtrsim 0.15\) [2510.13723]. In the asymmetric warm dark matter model, the asymmetry rather than annihilation is the dominant structural ingredient, but the same tiny Yukawa interaction that mixes the sterile fermion with active neutrinos also leaks a small amount of the dark asymmetry into the visible sector, providing baryogenesis by transfer rather than by freeze-out [2405.10303].

A frequent source of confusion is the assumption that ADM must annihilate into Standard Model states. The EFT reappraisal of Dirac-fermion ADM coupled directly to quarks or leptons shows why this is problematic: the annihilation required to erase the symmetric component imposes an upper bound on the EFT scale \(\Lambda\), while collider and direct-detection null results impose lower bounds. In that heavy-mediator EFT, quark-coupled ADM in the \(1\text{--}100~\mathrm{GeV}\) range is strongly constrained, whereas leptophilic ADM remains allowed for \(m_\chi \gtrsim 10~\mathrm{GeV}\) [2402.17265]. This suggests that hidden-sector annihilation channels are not an aesthetic addition but often the mechanism that preserves fermionic ADM viability.

## 4. Compact stars and fermionic ADM equations of state

Fermionic ADM enters compact-star physics through degenerate pressure and, in many models, through self-interactions encoded in a dark-sector equation of state. The generic scaling emphasized in both heuristic and GR treatments is
\[
M_{\max}\sim \frac{m_{\rm Pl}^3}{m_f^2},
\]
so lighter fermions can support more gravitating mass before relativistic instability sets in [1305.6908]. This is the central reason mixed stars can exceed the pure-neutron-star limit when the dark fermion is lighter than the neutron.

The GR two-fluid formulation is standard in this literature. For a mixed star composed of baryons and dark fermions, the stress tensor is written as
\[
T^{\mu  \nu} = T_1^{\mu  \nu}+ T_2^{\mu  \nu},
\]
with each fluid separately conserved, and the hydrostatic equations become two coupled TOV equations driven by the same total enclosed mass and total pressure [1305.6908]. In one mirror-like example with \(m_2=\tfrac12 m_n\), the authors find a mixed-star configuration with total mass \(3.74\,M_\odot\), neutron-sphere radius \(11.1\) km, total radius \(31.9\) km, and a dark mass component \(M_2=2.4\,M_\odot\) [1305.6908]. This is not a small-impurity scenario; it is a genuinely dark-matter-heavy compact object.

Self-interactions modify the stellar sequence in qualitatively different ways depending on whether they are repulsive or attractive. In the relativistic mean-field treatment of a Dirac spin-\(1/2\) ADM particle \(X\) with scalar attraction and vector repulsion,
\[
\mathcal L = \bar X\!\left[i\slashed{\partial}-g_V\slashed V-(m_X-g_\phi \phi)\right]X -\frac14 V_{\mu\nu}^2+\frac12 m_V^2 V_\mu^2 +\frac12(\partial_\mu\phi\,\partial^\mu\phi-m_\phi^2\phi^2)-V(\phi),
\]
the energy density and pressure are
\[
\epsilon = \frac{m_X^4}{3\pi^2} \left[ \frac{\varphi^2}{2C_\phi^2} +W(\varphi) +\frac{C_V^2}{2}\left(\frac{k_F}{m_X}\right)^6 +3\int_0^{k_F/m_X}x^2\sqrt{x^2+(1-\varphi)^2}\,dx \right],
\]
\[
p = \frac{m_X^4}{3\pi^2} \left[ -\frac{\varphi^2}{2C_\phi^2} -W(\varphi) +\frac{C_V^2}{2}\left(\frac{k_F}{m_X}\right)^6 +\int_0^{k_F/m_X}\frac{x^4\,dx}{\sqrt{x^2+(1-\varphi)^2}} \right].
\]
A major result is that strong scalar attraction increases, rather than decreases, the maximum stable ADM-star mass relative to free fermions; attractive self-interactions do not make fermionic ADM significantly better at destabilizing neutron stars than non-interacting fermionic ADM [1809.08254].

The hybrid-star study of self-interacting asymmetric dark matter adopts a simpler benchmark with \(m_\chi=1~\mathrm{GeV}\), vector-mediator mass \(m_I=100~\mathrm{MeV}\), and a repulsive Yukawa interaction added to the degenerate Fermi-gas equation of state:
\[
\varepsilon_{\chi int}=\frac{m_\chi}{\lambda_\chi^3} \chi(x_F) +\left(\frac{1}{3\pi^2}\right)^2 \frac{x_f^6m_\chi^6}{(\hbar c)^3m_I^2},
\]
\[
P_{\chi int}=\frac{m_\chi}{\lambda_\chi^3} \phi(x_F) +\left(\frac{1}{3\pi^2}\right)^2 \frac{x_f^6m_\chi^6}{(\hbar c)^3m_I^2}.
\]
Combined with a DDM3Y baryonic EoS and a two-fluid GR treatment, this yields pure self-interacting ADM stars with \(M_{\max}=3.0279\,M_\odot\), \(R=16.2349~\mathrm{km}\) in the static case and \(M_{\max}=3.1460\,M_\odot\), \(R_{eq}=19.2173~\mathrm{km}\) in the rotating case [1612.07093].

The same study also finds that the headline mixed-star configuration is dark-matter dominated. For fixed nuclear central enthalpy and varying dark central enthalpy, the maximum occurs for differential rotation with DM frequency \(700\) Hz and nuclear frequency \(300\) Hz:
\[
M=1.9355\,M_\odot,\qquad R=10.3717~\mathrm{km},
\]
with mass decomposition
\[
M_{nuc}=0.1179\,M_\odot,\qquad M_{dark}=1.9926\,M_\odot.
\]
The paper explicitly cautions that this is not an ordinary neutron star with a small ADM impurity; it is a dark-matter-dominated hybrid configuration [1612.07093]. That caution matters because the mass–radius point alone does not reveal the composition.

## 5. Neutron-star capture, ADM cores, and collapse bounds

Neutron stars probe fermionic ADM in two distinct regimes. One is the equilibrium-structure regime, where a dark core or mixed fluid changes global stellar properties. The other is the capture-to-collapse regime, where nonannihilating ADM accumulates over gigayear timescales and may form a black hole.

In the equilibrium-inference regime, Bayesian analyses that vary both the baryonic and ADM equations of state simultaneously find that currently available mass–radius data constrain only the combination
\[
\frac{g_\chi/m_\phi}{m_\chi},
\]
and only through a lower bound. Using NICER mass–radius data, the lower bound is
\[
\log_{10}\left( \frac{g_\chi/(m_\phi/\mathrm{MeV})}{m_\chi/\mathrm{MeV}} \right) \gtrsim -6.59 \quad \text{at 68\% credible level,}
\]
\[
\log_{10}\left( \frac{g_\chi/(m_\phi/\mathrm{MeV})}{m_\chi/\mathrm{MeV}} \right) \gtrsim -7.77 \quad \text{at 95\% credible level.}
\]
At the same time, \(m_\chi\), \(g_\chi\), and the ADM mass fraction \(F_\chi\) remain essentially unconstrained, and stars with \(F_\chi \le 1.7\%\) are nearly indistinguishable from purely baryonic stars for mass–radius uncertainties \(\ge 2\%\) [2410.00140]. This indicates that small fermionic ADM cores are observationally subtle in mass–radius space.

The collapse literature asks a different question: whether captured fermionic ADM can exceed its Chandrasekhar limit and form a black hole inside an old neutron star. A conservative TOV-based study had already argued that captured spin-\(1/2\) ADM—whether self-attractive, repulsive, or noninteracting—cannot destroy neutron stars over cosmic times unless the ADM mass is roughly PeV scale or larger, because the ADM core must itself exceed the maximum stable mass of an ADM star with the same equation of state [1809.08254]. In that treatment, the maximal capture fraction from the DM flux through the star gives
\[
m_X N_{X,\rm cap} \lesssim \left(3\times10^{-14}M_\odot\right) \frac{\rho_{\rm DM}}{\mathrm{GeV/cm^3}} \frac{200\,\mathrm{km/s}}{v_{\rm DM}} \frac{t}{10^{10}\,\mathrm{yr}},
\]
leading to \(m_{X,\rm collapse}\gtrsim 2\times10^5 - 2\times10^6\,\mathrm{GeV}\) depending on the local dark-matter density [1809.08254].

A more recent capture-to-collapse reassessment retains the same qualitative sequence—capture, thermalization, self-gravitation, Chandrasekhar collapse, black-hole formation, and competition between accretion and Hawking evaporation—but revises each step. The captured particle number is \(N_\chi(t)=Ct\), the thermalized cloud is modeled by
\[
n_\chi(r,t) \simeq \frac{N_\chi(t)}{\pi^{3/2} r_\chi^3(t)} \exp[-r^2/r_\chi^2(t)],
\]
the thermal radius is
\[
r_{\rm th} = \sqrt{\frac{3T}{\pi G(\rho_c+3P_c) m_\chi}},
\]
and the fermionic collapse threshold becomes
\[
N_\chi(t)> 2^{3/4} \pi \sqrt{3} \left(\frac{M_\text{Pl}}{m_\chi}\right)^3 \equiv N_\text{Ch}.
\]
Because the analysis uses a Gaussian ADM distribution rather than a uniform sphere, and because it incorporates realistic NS capture, thermalization, accretion, and evaporation, previous results are relaxed by a few orders of magnitude [2507.22881]. The revised bounds are relevant mainly for ultraheavy fermionic ADM, roughly
\[
10^7~{\rm GeV} \lesssim m_\chi \lesssim 10^{10}~{\rm GeV},
\]
rather than for canonical GeV-scale fermionic ADM [2507.22881].

A recurrent controversy concerns whether attractive self-interactions make fermionic ADM especially efficient at destabilizing neutron stars. The relativistic mean-field analysis states the opposite: attractive interactions soften the EoS only at lower densities, but by reducing the effective mass they accelerate the approach to the relativistic regime and do not substantially reduce the maximum stable ADM mass [1809.08254]. This controversy is therefore about the EoS treatment and the stability criterion, not merely about numerical details.

## 6. Detection channels, cosmological probes, and open disputes

The most constrained fermionic ADM scenarios are those in which the same operator both removes the symmetric component and mediates visible-sector scattering. In the heavy-mediator EFT with dimension-6 contact operators,
\[
\mathcal{O}_{\Gamma\Gamma'}=\frac{1}{\Lambda^2} (\bar{\psi}\Gamma\psi)(\bar{\chi}\Gamma'\chi),
\]
the practical ADM requirement is
\[
Y_{\rm sym}\leq \frac{1}{100}\times \frac{\Omega_{\rm DM} h^2}{2.76\times 10^8} \left(\frac{\rm GeV}{m_{\chi}}\right),
\]
so efficient annihilation imposes an upper bound on \(\Lambda\), while colliders and direct detection impose lower bounds [2402.17265]. In that framework, quark-coupled Dirac-fermion ADM is strongly constrained throughout the preferred \(1\text{--}100~\mathrm{GeV}\) range, while leptophilic ADM remains allowed for \(m_\chi \gtrsim 10~\mathrm{GeV}\), and FCC-ee could improve the reach to \(m_\chi\sim 175\text{--}200~\mathrm{GeV}\) and \(\Lambda \sim 1~\mathrm{TeV}\) [2402.17265]. The same study explicitly notes that excluding a region in \((m_\chi,\Lambda)\) does not exclude all dark matter of mass \(m_\chi\), because light mediators can substantially change annihilation, collider kinematics, and direct-detection rates [2402.17265].

Indirect detection can probe the asymmetry-transfer operator itself. For fermionic ADM operators of the form
\[
\frac{X u_i^c d_j^c d_k^c}{M_{ijk}^2},\qquad \frac{X q_i \ell_j d_k^c}{M_{ijk}^2},\qquad \frac{X \ell_i \ell_j e_k^c}{M_{ijk}^2},
\]
the same interaction that shares asymmetry between sectors also induces late decays, with lifetime
\[
c \tau \simeq \frac{6144 \pi^3 M^4}{C_{color} C_{flavor} C_{SU(2)_W} m_X^5}.
\]
A distinctive ADM feature is that only one of \(X\) or \(\bar X\) survives today, so the final-state charge asymmetry depends on the sign of the baryon or lepton number carried by the ADM particle [1401.7664]. The gamma-ray spectra are indifferent to that sign, but charged cosmic-ray signatures are not. Light ADM constraints correspond to \(M\sim 10^{13}\,\mathrm{GeV}\), while heavy ADM fits to AMS-02 and H.E.S.S. point to \(M\sim 10^{15}\text{--}10^{16}\,\mathrm{GeV}\), albeit still in tension with gamma rays and antiprotons [1401.7664].

Cosmological hidden-sector probes remain important precisely because many viable fermionic ADM models suppress visible signatures. If the mediator decays into dark radiation, precise \(N_{\rm eff}\) measurements and DM–DR scattering become central observables rather than side effects [1203.5803]. In unified Dirac-neutrino/ADM models, the same FN-like structure that suppresses unwanted right-handed-neutrino interactions yields
\[
\Delta N_{\mathrm{eff}} \lesssim 0.1,
\]
compatible with current bounds, while direct detection is loop-induced and future low-threshold experiments such as DS-LM-like searches can probe most of the remaining region [2510.13723].

At the opposite end of the mass spectrum, asymmetric warm dark matter predicts a soft X-ray signal with helicity asymmetry. The radiative decay
\[
\bar N\to \bar \nu_L +\gamma
\]
occurs at rate
\[
\Gamma_{N\rightarrow \nu_L\gamma} \approx 1.4\times 10^{-26} {\rm s}^{-1} \left(\frac{M_N}{1\,{\rm keV}}\right)^5 \left(\frac{\theta}{0.5\times 10^{-2}}\right)^2,
\]
and because the dark matter is asymmetric and chiral, the emitted photons are circularly polarized [2405.10303]. The paper describes helical X-rays as a smoking-gun signal of that scenario [2405.10303]. This is a reminder that fermionic ADM phenomenology is not confined to the canonical GeV-scale direct-detection paradigm.

Several disputes in the field are now comparatively well defined. One is whether neutron-star survival generically excludes fermionic ADM; current results indicate that this is model-dependent and typically weak except for ultraheavy ADM [1809.08254, 2507.22881]. Another is whether high-mass compact stars containing ADM should be interpreted as ordinary neutron stars with a small admixture; some explicit mixed-star solutions are instead dark-matter dominated and only mimic neutron-star masses and radii [1612.07093]. A third is whether present collider and direct-detection exclusions rule out fermionic ADM as such; the EFT answer is severe for heavy-mediator couplings to Standard Model fermions, but hidden-sector annihilation, light mediators, composite realizations, and dark-radiation cosmologies remain open [2402.17265, 1203.5803].

Fermionic ADM is therefore best understood not as a single model but as a research program organized around a common relic principle and a common microscopic ingredient: a non-self-conjugate or effectively non-self-conjugate fermion carrying a primordial asymmetry. Across that program, the central technical themes are the removal of the symmetric component, the preservation or late violation of dark number, the role of Fermi degeneracy in dense matter, and the fact that many observables constrain only effective parameter combinations rather than the underlying microphysics.

Source: https://www.emergentmind.com/topics/fermionic-asymmetric-dark-matter-adm