betaFeedback

WorksPhysics and Astronomy

Searching for Dark Photons and ALPs using Gamma-rays from Nuclear Reactors

Abstract

Nuclear reactors are intense sources of MeV gamma rays, produced in fission, radiative neutron capture and the de-excitation of fission products. Building on earlier phenomenological studies, we present a detector-independent description of the production of light, weakly coupled particles by reactor gamma rays and of their detection through scattering, absorption and decay in a nearby detector. We consider kinetically mixed dark photons, produced and detected through Compton-like processes, and axion-like particles (ALPs) coupled to photons, electrons and nucleons, produced through Primakoff conversion, Compton-like scattering and magnetic nuclear transitions such as the 2.23 MeV2.23~\mathrm{MeV} line of p(n,γ)dp(n,γ)d. As a concrete application, we derive the projected sensitivities of the CONUS+ germanium experiment at the Leibstadt reactor and of a conceivable 100100-kg CONUS100 upgrade. For sub-MeV dark photons, CONUS+ and CONUS100 reach kinetic mixings of ε≃1.65×10−5ε\simeq1.65\times10^{-5} and 1.2×10−51.2\times10^{-5}, respectively, for 0.5 keV≲mA′≲0.5 MeV0.5~\mathrm{keV}\lesssim m_{A'}\lesssim0.5~\mathrm{MeV}. For ALPs, we show that CONUS100 can probe an unexplored region of parameter space at MeV masses, in particular through a→e+e−a\to e^+e^- decays of ALPs emitted in the deuteron line, assuming a energy window extended to 10 MeV10~\mathrm{MeV}.

Introduction

A wide range of astrophysical and cosmological observations indicates that most of the matter content of the Universe is non-luminous and cannot be accounted for within the Standard Model (SM) of particle physics. Evidence for dark matter is inferred from gravitational phenomena over very different length scales, including galactic rotation curves, the dynamics and gravitational lensing of galaxy clusters, and the anisotropies of the cosmic microwave background [1]. Despite the robustness of this gravitational evidence, the microscopic nature of dark matter remains unknown.

For several decades, weakly interacting massive particles (WIMPs) provided one of the main benchmarks for experimental dark-matter searches. Their connection to physics at the electroweak scale and the possibility of obtaining the observed relic abundance through thermal freeze-out motivated an extensive experimental program based on direct detection, indirect searches, and collider experiments [2, 3, 4]. The absence of a conclusive WIMP signal, together with increasingly stringent experimental constraints, has motivated a substantial broadening of the search program beyond the traditional weak-scale paradigm [5].

Particular attention has consequently been given to dark sectors containing new particles at sub-GeV and MeV scales. Such sectors may contain light dark matter states as well as new mediators connecting otherwise secluded degrees of freedom to the SM. Two well-motivated examples are dark photons and axion-like particles (ALPs). A dark photon, associated with an additional Abelian gauge symmetry, can interact with electrically charged SM particles through kinetic mixing with the ordinary photon [6, 7]. ALPs instead arise naturally as light pseudoscalar degrees of freedom and can interact with photons, electrons, and nucleons through a variety of effective operators [8, 9, 10]. Depending on the underlying theory, these particles can themselves participate in the dark-matter sector or provide light portals through which a hidden sector can be experimentally probed.

The search for light and weakly coupled states poses a distinct experimental challenge. As the relevant masses and momentum transfers decrease, their signals can fall below the thresholds of conventional high-energy searches, while their small couplings imply correspondingly suppressed production and interaction rates. These considerations motivate experiments that combine intense particle sources, short source - detector distances, low detection thresholds, and well-controlled backgrounds. Nuclear reactors provide an interesting environment in this respect. In addition to their intense antineutrino emission, reactor cores generate a large flux of photons with energies extending into the MeV range through fission, radiative capture, and the de-excitation of fission products and other nuclei [10, 11, 12]. These photons, as well as specific magnetic nuclear transitions in the reactor environment, can act as sources of light weakly coupled particles.

Reactor experiments can therefore provide laboratory probes of dark-sector states through production and detection mechanisms that are complementary to those explored in conventional dark matter searches. Reactor photons can produce dark photons through Compton-like conversion on electrons, while ALPs can be generated through Primakoff conversion, Compton-like scattering, or magnetic nuclear de-excitation, depending on their dominant coupling [7, 10, 13]. Once produced, these particles may propagate out of the reactor core and generate observable signatures in a nearby detector through scattering, absorption, or decay. The combination of an intense source with a short baseline can be particularly advantageous for particles whose interactions are too weak to be efficiently tested in smaller-scale laboratory sources.

Reactor searches for axions emitted in magnetic nuclear transitions, motivated by the original arguments of Refs. [14, 15], date back to the Bugey experiment [16], which was later extended by TEXONO [17]. More recently, the NEON experiment has set new constraints on ALPs at a nuclear reactor as is shown in [18]. A different search was proposed in Ref. [13] based on ALP scattering at reactor neutrino experiments. See Ref. [10] for a systematic study of ALPs at reactor experiments.

In this work we derive the projected sensitivities of CONUS+, a germanium experiment at 20.7 m from the 3.6 GW core of the Leibstadt reactor [19], and of a conceivable 100-kg CONUS100 upgrade [20], taking into account their detector geometry and a per-fission normalization of the nuclear-line yields, and compare them with current experimental limits.

The paper is organized as follows. The first part is independent of a specific detector. In § 2 we describe reactor gamma rays as a source of light weakly coupled particles, both through the continuous photon spectrum and through discrete magnetic nuclear transitions. § 3 introduces the dark photon scenario and its production and detection mechanisms, and in § 4 we present the ALP framework and the production and detection channels associated with photon, electron, and nucleon couplings. In § 5 we apply this framework to CONUS+ and to the CONUS100 benchmark: we describe the experimental setup and the effective detector geometry, show representative event rates, and discuss the projected sensitivities for the different dark photon and ALP scenarios in § 5.5. Finally, § 6 summarizes our main results and conclusions.

Reactor gamma rays as a source of light dark-sector particles

Continuous reactor photon spectrum

The production of weakly coupled particles in a nuclear reactor is driven by the intense prompt photon emission generated by fission fragments and their subsequent de-excitations. We model the continuous reactor photon spectrum through the empirical parameterization [10, 12],

dNγdEγ=5.8×1017(PMW)exp⁡[−1.1EγMeV] MeV−1s−1,(1)\frac{\mathrm{d}N_{\gamma}}{\mathrm{d}E_{\gamma}} = 5.8 \times10^{17} \left(\frac{P}{\mathrm{MW}}\right) \exp\left[-1.1\frac{E_{\gamma}}{\mathrm{MeV}}\right] \,\mathrm{MeV}^{-1}\mathrm{s}^{-1}, \tag*{(1)}

where PP is the reactor thermal power. Although this quantity is often referred to as the reactor photon flux, its units show that it represents the total differential photon emission rate of the core, rather than a flux per unit area. The flux at a detector at a distance LL from the core is obtained by dividing the production rate of the new particle by 4πL24\pi L^{2}.

In our case, Eq. (1) is evaluated up to Emax⁡=10E_{\max} = 10 MeV. The lower limit depends on the process: for example, 25 keV for ALP production, as in Ref. [10], and 0.2 MeV for dark photon production, as in Ref. [7]. The boundaries define the theoretical integration window and should not be identified with the instrumental energy acceptance of a specific detector. Equation (1) is an empirical parameterization, not the result of a photon transport simulation of a specific core. We assume that the photon emission scales linearly with the reactor thermal power and neglect variations associated with fuel evolution, core geometry, structural materials and reactor operation; the core is treated as a pointlike, isotropic source. For a reactor experiment such as CONUS+, at the 3.6 GW KKL core (Table 2), the continuous spectrum is simply evaluated with P=3.6P = 3.6 GW. The nuclear-line emission is instead taken from TEXONO, at a 2.9 GW reactor [17], and scaled to KKL through the fission rate of Eq. (2), i.e. by the ratio 3.6/2.9; the resulting rates are shown on the right axis of Figure 1.

Figure 1. Photon yield per fission, Φγ,i\Phi_{\gamma,i}, of the six magnetic nuclear transitions, taken from TEXONO [10, 17] (left axis), and the corresponding emission rate at the 3.6 GW KKL core, Rf(P)Φγ,iR_f(P)\Phi_{\gamma,i}, with RfR_f from Eq. (2) (right axis).

ParameterCONUS+CONUS100
Reactor thermal power PP3.6 GW3.5 GW
Reactor–detector distance LL20.7 m20 m
Ge target mass MdetM_{\mathrm{det}}3.74 kg100 kg
Exposure Δt\Delta t1 FPE yr5 FPE yr
Accepted energy window25 keV (0.2 MeV)–0.6 MeV25 keV (0.2 MeV)–10 MeV

Table 2. Experimental benchmarks adopted for the dark photon and ALPs projections [19, 20]. The CONUS+ entries follow the present detector configuration, while the CONUS100 entries correspond to the proposed 100 kg benchmark. The lower edge of the energy window is 25 keV for ALPs and 0.2 MeV (in parentheses) for dark photons.

This continuous photon spectrum is used below for dark photon production and for ALP production through Primakoff conversion and Compton-like scattering. ALP production through nuclear de-excitation is treated separately because it is determined by the discrete photon yields of specific magnetic nuclear transitions rather than by the continuous spectrum of Eq. (1).

Discrete magnetic nuclear transitions

Besides the continuous spectrum, the reactor environment emits monochromatic photons in specific nuclear transitions. Magnetic (MJMJ) transitions are of particular interest, because a pseudoscalar particle coupled to nucleons can be emitted instead of the photon (§ 4.2.3). We adopt the six magnetic transitions used in the TEXONO analysis [17] and in Ref. [10]. Thermal-neutron capture on hydrogen in the reactor coolant, p(n,γ)dp(n,\gamma)d, produces a 2.230 MeV line that is predominantly an isovector M1M1 transition. The 10B(n,α)7Li∗{}^{10}\mathrm{B}(n,\alpha){}^{7}\mathrm{Li}^{*} reaction populates the first excited state of 7Li{}^{7}\mathrm{Li}, whose predominantly M1M1 de-excitation produces a 478 keV line. The remaining four sources are magnetic M4M4 transitions in the fission fragments 91Y∗{}^{91}\mathrm{Y}^{*} (555 keV), 97Nb∗{}^{97}\mathrm{Nb}^{*} (743 keV), 135Xe∗{}^{135}\mathrm{Xe}^{*} (526 keV), and 137Ba∗{}^{137}\mathrm{Ba}^{*} (662 keV). The neutron-capture line and the four M4M4 lines were considered in the Bugey search [16]; TEXONO subsequently included the 7Li∗{}^{7}\mathrm{Li}^{*} transition [17], using nuclear-structure calculations from Refs. [17, 21].

The photon-line yields per fission adopted in the calculation are summarized in Figure 1. We denote the yield of line ii by Φγ,i\Phi_{\gamma,i}, which is dimensionless when quoted as photons per fission. The TEXONO per-fission yields [10, 17] are adopted as benchmark values for the reactor configurations considered here; differences in fuel evolution, control-rod composition, neutron transport and core geometry are neglected, so these yields are an approximate reactor model rather than the result of a dedicated KKL core simulation. Because it is a per-fission quantity, Φγ,i\Phi_{\gamma,i} is not multiplied by a reactor-power ratio. Reactor power enters only through the fission rate defined below. This prescription avoids counting the power dependence twice: since Rf∝PR_f \propto P, it is equivalent to the single rescaling by the power ratio described in § 2.1, Rf(P)Φγ,i=(P/PTEXONO)Rf(PTEXONO)Φγ,iR_f(P)\Phi_{\gamma,i}=(P/P_{\mathrm{TEXONO}})R_f(P_{\mathrm{TEXONO}})\Phi_{\gamma,i}. It differs from Ref. [10], where the TEXONO yields were additionally rescaled by 4/2.9 (the ratio of reactor powers); for the same reactor this would increase the line rates by 38% and strengthen the projected bounds on ∣gaγγ(1)∣\lvert g_{a\gamma\gamma}^{(1)}\rvert and ∣gaee(1)∣\lvert g_{aee}^{(1)}\rvert by a factor ≃1.17\simeq1.17.

The fission rate is normalized following Ref. [10] as

Rf(P)=1.248×1020(P4 GW)s−1,(2)R_f(P)=1.248\times10^{20}\left(\frac{P}{4\ \mathrm{GW}}\right)\mathrm{s}^{-1}, \tag*{(2)}

so that the emission rate of line ii is Rf(P)Φγ,iR_f(P)\Phi_{\gamma,i}. This simple conversion assumes that the fission rate scales linearly with thermal power; the normalization corresponds to an energy release of 200 MeV per fission, 1 GW/200 MeV≃1.248×1020 s−11\ \mathrm{GW}/200\ \mathrm{MeV}\simeq1.248\times10^{20}\ \mathrm{s}^{-1}.

Dark Photon

The model

We consider a minimal dark photon scenario in which the SM electromagnetic sector is supplemented by an additional Abelian gauge field Aμ′A^{\prime}_{\mu} associated with a hidden U(1)D\mathrm{U}(1)_{\mathrm{D}} symmetry. At low energies, the visible and dark gauge sectors can communicate through kinetic mixing,

L=−14FμνFμν−14Fμν′F′μν−ϵ2FμνF′μν+12mA′2Aμ′A′μ−eAμJEMμ.(3)\mathcal{L}=-\frac{1}{4}F_{\mu\nu}F^{\mu\nu}-\frac{1}{4}F^{\prime}_{\mu\nu}F^{\prime\mu\nu}-\frac{\epsilon}{2}F_{\mu\nu}F^{\prime\mu\nu}+\frac{1}{2}m_{A^{\prime}}^{2}A^{\prime}_{\mu}A^{\prime\mu}-eA_{\mu}J_{\mathrm{EM}}^{\mu}. \tag*{(3)}

where FμνF_{\mu\nu} and Fμν′F^{\prime}_{\mu\nu} are the electromagnetic and dark photon field strength tensors, respectively, mA′m_{A^{\prime}} denotes the dark photon mass, and ϵ\epsilon is the kinetic-mixing parameter. After diagonalizing and canonically normalizing the gauge kinetic terms, the dark photon acquires an effective coupling to the electromagnetic current. To leading order in ϵ\epsilon, the interaction relevant for the present analysis can be written as

Lint=−ϵeAμ′JEMμ.(4)\mathcal{L}_{\mathrm{int}}=-\epsilon e A^{\prime}_{\mu}J_{\mathrm{EM}}^{\mu}. \tag*{(4)}

In particular, the coupling to electrons is

LA′e=−ϵeAμ′eˉγμe.(5)\mathcal{L}_{A^{\prime}e}=-\epsilon e A^{\prime}_{\mu}\bar{e}\gamma^{\mu}e. \tag*{(5)}

Therefore, the phenomenology considered here is fully specified by the two independent parameters: mA′m_{A'} and ϵ\epsilon.

The interaction in Eq. (5) allows reactor photons to produce dark photons through Compton-like conversion on electrons, while dark photons reaching the germanium detector can be observed through the crossed process (Figure 2). Since the same effective interaction controls production and detection, both cross sections are proportional to ϵ2\epsilon^2, and the event yield scales approximately as ϵ4\epsilon^4 in the absence of propagation effects.

Dark photon production and detection processes

Figure 2. Dark photon production and detection processes. (a) Production via Compton-like scattering γ+e−→A′+e−\gamma+ e^- \to A' + e^-. (b) Detection via the inverse process, A′+e−→γ+e−A' + e^- \to\gamma+ e^-.

In this work, we restrict the analysis to mA′<2mem_{A'} < 2m_e, so that the tree-level decay A′→e+e−A' \to e^+e^- is kinematically forbidden. We also consider only mA′≥100m_{A'} \geq100 eV: in a medium the photon acquires an effective mass equal to the plasma frequency, mγ≃20m_\gamma\simeq20 eV in the reactor and ≃44\simeq44 eV in germanium, and for mA′≲mγm_{A'} \lesssim m_\gamma both the production and the detection probabilities are suppressed by (mA′/mγ)4(m_{A'}/m_\gamma)^4 [22], while for mA′≳100m_{A'} \gtrsim100 eV this effect changes ϵ\epsilon by less than ∼15%\sim15\%. In the minimal visible-sector scenario considered here, the remaining loop induced visible decays are sufficiently suppressed over the parameter region of interest and are neglected. We also assume that no lighter dark-sector states are available as invisible decay products. The production and detection rates associated with these processes are discussed in the following subsection.

Production and detection mechanism

Production

The production channel under consideration is the inelastic scattering of a reactor photon off an atomic electron, γ+e−→A′+e−\gamma+ e^- \to A' + e^- (see Figure 2a), where A′A' denotes the massive dark photon. We use the exact tree-level cross section for a massive vector, in which the dark photon mass is kept in the matrix element, in the kinematics and in the production threshold ω≥mA′+mA′2/(2me)\omega\geq m_{A'} + m_{A'}^2/(2m_e), without assuming mA′≪ωm_{A'} \ll\omega. For a dark photon mass mA′≪mem_{A'} \ll m_e (where me=0.511m_e = 0.511 MeV), it reduces to the ordinary Klein-Nishina differential cross-section evaluated at the same outgoing energy:

dσγ→A′dEA′(ω,EA′)≈ϵ2dσKNdω′(ω,ω′)∣ω′=EA′(6)\frac{d\sigma_{\gamma\to A'}}{dE_{A'}}(\omega,E_{A'}) \approx\epsilon^2 \left.\frac{d\sigma_{\mathrm{KN}}}{d\omega'}(\omega,\omega')\right|_{\omega'=E_{A'}} \tag*{(6)}

where EA′E_{A'} is the total energy of the produced dark photon. [7] replaced the total photon cross section of the reactor medium by the Klein-Nishina one. Below ∼1\sim1 MeV, however, photoelectric absorption is important in uranium, and the Compton fraction of the total cross section is only 11–42% between 0.25 and 0.6 MeV. We therefore use the total photon cross section of uranium from XCOM [23], σtot(ω)\sigma_{\mathrm{tot}}(\omega), and multiply the cross section per electron of Eq. (6) by ZU=92Z_U = 92.

The differential production rate of dark photons in the reactor core is obtained by folding the conversion probability with the source spectrum:

FA′(EA′)≡dN˙A′dEA′=∫ωmin⁡ωmax⁡dωZUσtotU(ω)dσγ→A′dEA′(ω,EA′)dN˙γdω,(7)\mathcal{F}_{A'}(E_{A'}) \equiv\frac{\mathrm{d}\dot{N}_{A'}}{\mathrm{d}E_{A'}} = \int_{\omega_{\min}}^{\omega_{\max}} \mathrm{d}\omega\frac{Z_U}{\sigma_{\mathrm{tot}}^{\mathrm{U}}(\omega)} \frac{\mathrm{d}\sigma_{\gamma\to A'}}{\mathrm{d}E_{A'}}(\omega,E_{A'}) \frac{\mathrm{d}\dot{N}_{\gamma}}{\mathrm{d}\omega}, \tag*{(7)}

where dN˙γ/dω\mathrm{d}\dot{N}_{\gamma}/\mathrm{d}\omega is the photon emission rate of Eq. (1) and the integration limits are implicitly constrained by the kinematic requirement that the outgoing dark photon energy must lie within the physical Compton continuum for the given incident photon energy. Equation (7) is evaluated numerically with incident photon energies spanning ω∈[0.2,10]\omega\in[0.2, 10] MeV, the lower edge being the one adopted for the parameterization of Eq. (1) in Ref. [7].

Two refinements of this treatment have been discussed in the literature. In the oscillation description of Ref. [22], valid for mγ≪mA′≪ωm_{\gamma} \ll m_{A'} \ll\omega, every reactor photon converts with probability ϵ2\epsilon^{2} independently of the absorption in the medium, and only the two transverse polarizations are produced. In addition, Compton-scattered photons produce further dark photons [24]; Ref. [25] finds this to increase the flux by a factor 1.2–2.6 between 2 and 0.2 MeV. For our CONUS+ and CONUS100 benchmarks these treatments would lower the projected ϵ\epsilon by about 30–35% for mA′≪ωm_{A'} \ll\omega, where they apply. The calculation adopted here, which keeps the full mass dependence needed for mA′∼ωm_{A'} \sim\omega, is therefore conservative.

Detection

Once produced, the dark photons propagate to the detector with negligible attenuation, as the decay length far exceeds reactor–detector baselines of a few tens of metres. Detection proceeds via the time-reversed process, A′+e−→γ+e−A' + e^{-} \to\gamma+ e^{-} (see Figure 2b). The integrated detection cross-section for a given dark photon energy EA′E_{A'} is

ΣA′(EA′)→mA′→023ϵ2∫Emin⁡Emax⁡dER dσKNdER(ω=EA′,ER),(8)\Sigma_{A'}(E_{A'}) \xrightarrow{m_{A'} \to0} \frac{2}{3}\epsilon^{2}\int_{E_{\min}}^{E_{\max}} \mathrm{d}E_{R}\,\frac{\mathrm{d}\sigma_{\mathrm{KN}}}{\mathrm{d}E_{R}}(\omega= E_{A'}, E_{R}), \tag*{(8)}

where ERE_{R} denotes the energy of the recoil photon produced in the final state. The factor 2/3 originates from the averaging over the three helicity/polarisation states of the massive dark photon. For finite mA′m_{A'} we obtain ΣA′\Sigma_{A'} from the exact production cross section by detailed balance, which keeps the 2/3 factor and the ratio of the phase-space momenta. For mA′≪EA′m_{A'} \ll E_{A'}, where only transverse dark photons are produced [22], the 2/3 factor makes the detection rate conservative by a factor 3/2. In the massless limit the kinematic boundaries of the integral are determined by the Compton edge:

Emin⁡(EA′)=EA′1+2EA′/me,Emax⁡(EA′)=EA′.(9)E_{\min}(E_{A'}) = \frac{E_{A'}}{1 + 2E_{A'}/m_{e}}, \qquad E_{\max}(E_{A'}) = E_{A'}. \tag*{(9)}

For the purposes of computing the differential event rate, the integration is performed over the entire kinematically accessible range, corresponding to a detector with full acceptance and zero energy threshold. Assuming that the final-state photon and electron are fully absorbed, the deposited energy is Edep=EA′E_{\mathrm{dep}} = E_{A'}, and the number of events in an accepted window [Edepmin⁡,Edepmax⁡][E_{\mathrm{dep}}^{\min}, E_{\mathrm{dep}}^{\max}] is

Nsig=Ne Δt4πL2∫max⁡(Edepmin⁡,mA′)Edepmax⁡dEA′ FA′(EA′) ΣA′(EA′),(10)N_{\mathrm{sig}} = \frac{N_{e}\,\Delta t}{4\pi L^{2}}\int_{\max(E_{\mathrm{dep}}^{\min},m_{A'})}^{E_{\mathrm{dep}}^{\max}} \mathrm{d}E_{A'}\,F_{A'}(E_{A'})\,\Sigma_{A'}(E_{A'}), \tag*{(10)}

where NeN_{e} is the number of electrons in the germanium target, Δt\Delta t the exposure and LL the reactor–detector distance. The lower limit mA′m_{A'} is reached only for mA′<mem_{A'} < m_{e}: for heavier dark photons the kinematics of γe−→A′e−\gamma e^{-} \to A'e^{-} on an electron at rest requires EA′≥(mA′2+me2)/(2me)E_{A'} \ge(m_{A'}^{2} + m_{e}^{2})/(2m_{e}), e.g. 1.23 MeV for mA′=1m_{A'} = 1 MeV, so the spectra of Figure 3 and Figure 11 start above EA′=mA′E_{A'} = m_{A'}. Since production and detection each contribute ϵ2\epsilon^{2},

Dark photon production spectrum at the KKL core for three values of $\epsilon$ and three dark-photon masses

Figure 3. dark photon production spectrum at the KKL core (3.6 GW) for ϵ=1\epsilon= 1 (top left), 10−310^{-3} (top right) and 10−610^{-6} (bottom), and mA′=0.1m_{A'} = 0.1 (blue), 0.5 (black) and 1 MeV (green), from reactor photons above 0.2 MeV (the shaded area is below this energy).

Differential dark photon event rates for CONUS+ and CONUS100

Figure 11. Differential dark photon event rates, dNobs/dEA′dN_{\mathrm{obs}}/dE_{A'}, per MeV and per full-power year, from inverse Compton-like absorption A′e−→γe−A'e^{-} \to\gamma e^{-} in CONUS+ (left) and CONUS100 (right) for mA′=0.1m_{A'} = 0.1, 0.50.5 and 11 MeV (rows) and ϵ=10−4\epsilon= 10^{-4}, 10−510^{-5} and 10−610^{-6} (colours). No efficiency or energy window is applied; the shaded areas are outside the accepted windows, below 0.20.2 MeV and, for CONUS+, above 0.60.6 MeV. The dotted lines mark EA′=mA′E_{A'} = m_{A'}.

Nsig=ϵ4N0(mA′),(11)N_{\mathrm{sig}} = \epsilon^{4}N_{0}(m_{A'}), \tag*{(11)}

where N0(mA′)N_{0}(m_{A'}) is the yield evaluated at $\epsilon = 1.

Axion-like Particles

Model

We consider a generic axion-like particle (ALP), denoted by aa, coupled to photons, electrons, and nucleons. At the energies relevant for reactor experiments, the interactions are parametrized by the effective Lagrangian [8, 10]

L=La−14gaγγaFμνF~μν−igaeeaeˉγ5e−iaNˉγ5(gann(0)+τ3gann(1))N,(12)\mathcal{L} = \mathcal{L}_a - \frac{1}{4}g_{a\gamma\gamma}aF_{\mu\nu}\widetilde{F}^{\mu\nu} - ig_{aee}a\bar{e}\gamma_5e - ia\bar{N}\gamma_5\left(g_{ann}^{(0)}+\tau_3g_{ann}^{(1)}\right)N, \tag*{(12)}

where

La=12∂μa∂μa−12ma2a2(13)\mathcal{L}_a = \frac{1}{2}\partial_\mu a\partial^\mu a-\frac{1}{2}m_a^2a^2 \tag*{(13)}

contains the free ALP kinetic and mass terms. Here FμνF_{\mu\nu} is the electromagnetic field-strength tensor, F~μν=ϵμνρσFρσ/2\widetilde{F}^{\mu\nu}=\epsilon^{\mu\nu\rho\sigma}F_{\rho\sigma}/2 is its dual, N=(p,n)TN=(p,n)^T denotes the proton–neutron isospin doublet, and τ3\tau_3 is the third Pauli matrix in isospin space. The couplings gann(0)g_{ann}^{(0)} and gann(1)g_{ann}^{(1)} are the isoscalar and isovector ALP–nucleon couplings, respectively.

The photon coupling gaγγg_{a\gamma\gamma} has dimensions of inverse energy, whereas the pseudoscalar electron and nucleon couplings in Eq. (12) are dimensionless. The fermionic terms can be related to the derivative ALP–fermion interaction [8, 9, 10]

Laf=Caf2fafˉγμγ5f ∂μa,f=e,p,n,(14)\mathcal{L}_{af}=\frac{C_{af}}{2f_a}\bar{f}\gamma^\mu\gamma_5f\,\partial_\mu a,\qquad f=e,p,n, \tag*{(14)}

which, using the fermion equations of motion, gives

gaff=mfCaffa.(15)g_{aff}=\frac{m_fC_{af}}{f_a}. \tag*{(15)}

Similarly, the photon coupling can be written schematically as

gaγγ=αCaγ2πfa,(16)g_{a\gamma\gamma}=\frac{\alpha C_{a\gamma}}{2\pi f_a}, \tag*{(16)}

where mfm_f is the fermion mass, α\alpha is the electromagnetic fine-structure constant, faf_a denotes the characteristic ALP interaction scale, and CafC_{af} and CaγC_{a\gamma} are model-dependent dimensionless coefficients.

We adopt a phenomenological ALP framework rather than a specific ultraviolet realization. The ALP mass and the effective couplings are therefore treated as independent parameters, and no QCD-axion mass–coupling relation is imposed. This allows the photon-, electron-, and nucleon-mediated reactor signatures to be analyzed separately before considering the mixed-coupling cases associated with nuclear production.

Production and detection mechanisms

Photon–ALP coupling

We first consider the single-coupling scenario in which gaγγg_{a\gamma\gamma} is non-zero, while the electron and nucleon couplings in Eq. (12) are set to zero. Reactor photons then produce ALPs through Primakoff conversion, γ+N→a+N\gamma+N\to a+N, while detection in germanium proceeds through inverse Primakoff scattering, a+N→γ+Na + N \to\gamma+ N, or through a→γγa \to\gamma\gamma decays occurring inside the effective detector volume. Loop induced interactions and decay modes not associated with gaγγg_{a\gamma\gamma} are neglected in this single coupling benchmark. The calculation follows the reactor treatment of Ref. [10], with the reactor source of § 2; the experimental benchmarks are specified in § 5.

ALP production and detection processes through the photon-ALP coupling

Figure 4. ALP production and detection processes through the photon-ALP coupling gaγγg_{a\gamma\gamma}. (a) ALP production via Primakoff conversion γ+N→a+N\gamma+ N \to a + N. (b) and (c) are the detection mechanisms considered: (b) inverse Primakoff scattering a+N→γ+Na + N \to\gamma+ N and (c) ALP decay into a photon pair, a→γγa \to\gamma\gamma.

Reactor production and propagation We use the reactor photon source of § 2.1 and, as in Ref. [10], an effective 235U{}^{235}\mathrm{U} production medium with ZU=92Z_U = 92, AU=235.04A_U = 235.04 and MU=AU×931.5 MeVM_U = A_U \times931.5\ \mathrm{MeV}. To distinguish the incident reactor photon from photons produced in the detector, we denote its energy by Eγ′E'_\gamma.

For a target nucleus with mass MNM_N and atomic number ZZ, the differential Primakoff production cross section is taken from the exact expression of Refs. [10, 26]

dσPdt=2αZ2F2(t)gaγγ2MN4t2(MN2−s)2(t−4MN2)2{ma2t(MN2+s)−ma4MN2−t[(MN2−s)2+st]}(17)\frac{\mathrm{d}\sigma_P}{\mathrm{d}t} = \frac{2\alpha Z^2 F^2(t)g_{a\gamma\gamma}^2 M_N^4}{t^2(M_N^2-s)^2(t-4M_N^2)^2}\left\{m_a^2t(M_N^2+s)-m_a^4M_N^2-t\left[(M_N^2-s)^2+st\right]\right\} \tag*{(17)}

where

s=MN2+2MNEγ′(18)s = M_N^2 + 2M_NE'_\gamma \tag*{(18)}

and t<0t < 0 is the Mandelstam variable associated with the transfer of momentum. Equation (17) does not rely on either the forward-scattering or the massless-ALP approximation. The total production cross section is obtained by integrating over the kinematically allowed momentum transfer,

σPP(Eγ′,N)=∫tmin⁡tmax⁡dσPPdt dt(19)\sigma_{\mathrm{P}}^{\mathrm{P}}(E'_{\gamma},N)=\int_{t_{\min}}^{t_{\max}}\frac{\mathrm{d}\sigma_{\mathrm{P}}^{\mathrm{P}}}{\mathrm{d}t}\,\mathrm{d}t \tag*{(19)}

with the exact two-body limits used in Refs. [10, 27],

tmin⁡=ma44s−(pcm+kcm)2,tmax⁡=ma44s−(pcm−kcm)2,(20)t_{\min}=\frac{m_a^4}{4s}-(p_{\mathrm{cm}}+k_{\mathrm{cm}})^2,\qquad t_{\max}=\frac{m_a^4}{4s}-(p_{\mathrm{cm}}-k_{\mathrm{cm}})^2, \tag*{(20)}
pcm=s−MN22s,kcm=[(s+ma2−MN2)24s−ma2]1/2.(21)p_{\mathrm{cm}}=\frac{s-M_N^2}{2\sqrt{s}},\qquad k_{\mathrm{cm}}=\left[\frac{(s+m_a^2-M_N^2)^2}{4s}-m_a^2\right]^{1/2}. \tag*{(21)}

The reaction is kinematically allowed only when s≥(MN+ma)2s\geq(M_N+m_a)^2, which imposes the incident-photon threshold

Eγ,thr′=ma+ma22MN.(22)E'_{\gamma,\mathrm{thr}}=m_a+\frac{m_a^2}{2M_N}. \tag*{(22)}

Together with the lower boundary of the reactor photon spectrum, this defines the integration limit

Elo(ma)=max⁡(Emin⁡,Eγ,thr′).(23)E_{\mathrm{lo}}(m_a)=\max(E_{\min},E'_{\gamma,\mathrm{thr}}). \tag*{(23)}

Atomic screening and the loss of nuclear coherence at finite momentum transfer are incorporated through a piecewise form factor. Defining q≡−tq\equiv\sqrt{-t}, for −t≤7.39me2-t\leq7.39m_e^2, where the momentum transfer is small enough for the atomic electrons to screen the nuclear charge [28], we use the atomic form factor of the Thomas–Fermi atom in the Molière representation [29], written for t<0t<0 as in Ref. [10],

FA2(t)=(a02t1−a02t)2,a0=184.15exp⁡(1/2)Z1/3me.(24)F_A^2(t)=\left(\frac{a_0^2t}{1-a_0^2t}\right)^2,\qquad a_0=\frac{184.15}{\exp(1/2)Z^{1/3}m_e}. \tag*{(24)}

For larger momentum transfer, it is replaced by the Helm nuclear form factor [30], as in Ref. [10] (Ref. [26] uses instead a two-parameter Fermi form factor; the difference is negligible here [31]),

FH2(q)=[3j1(qR1)qR1]2e−(qsH)2,R1=R2−5sH2,R=1.2A1/3 fm,sH=0.9 fm,(25)F_H^2(q)=\left[\frac{3j_1(qR_1)}{qR_1}\right]^2e^{-(qs_H)^2},\qquad R_1=\sqrt{R^2-5s_H^2},\qquad R=1.2A^{1/3}\,\mathrm{fm},\qquad s_H=0.9\,\mathrm{fm}, \tag*{(25)}

where j1j_1 denotes the spherical Bessel function of the first kind and AA is the nuclear mass number. The atomic-screening correction is particularly important in the low-mass region, ma≲104 eVm_a\lesssim10^4\,\mathrm{eV}, where Primakoff production is dominated by small momentum transfers. The Helm correction has a smaller numerical effect over most of the parameter space considered here.

Having specified the Primakoff production cross section, the corresponding photon-to-ALP conversion probability must be normalized to all competing photon interactions in the reactor fuel. We obtain the Standard Model photon cross section from the “total with coherent scattering” mass attenuation coefficient for uranium tabulated in the XCOM database [23]. This quantity includes coherent and incoherent scattering, photoelectric absorption, and electron-positron pair production whenever kinematically allowed. The XCOM mass attenuation coefficient is converted from cm2 g−1\mathrm{cm}^2\,\mathrm{g}^{-1} to a cross section per uranium atom according to

σSMU(E)=mmolUNA(μρ)XCOM(E),(26)\sigma_{\mathrm{SM}}^{U}(E)=\frac{m_{\mathrm{mol}}^{U}}{N_A}\left(\frac{\mu}{\rho}\right)_{\mathrm{XCOM}}(E), \tag*{(26)}

where mmolUm_{\mathrm{mol}}^{U} is the molar mass of natural uranium and NAN_A is Avogadro’s constant. The total interaction cross section entering the conversion probability is therefore

σTotU(E)=σSMU(E)+σPU(E;U).(27)\sigma_{\mathrm{Tot}}^{U}(E)=\sigma_{\mathrm{SM}}^{U}(E)+\sigma_{\mathrm{P}}^{U}(E;U). \tag*{(27)}

The probability for a photon of energy EE to produce an ALP is consequently σPU/σTotU\sigma_{\mathrm{P}}^{U}/\sigma_{\mathrm{Tot}}^{U}. Retaining the Primakoff contribution in the denominator guarantees that this probability remains smaller than unity, although in the coupling region relevant for the projected sensitivity one typically has σPU≪σSMU\sigma_{\mathrm{P}}^{U}\ll\sigma_{\mathrm{SM}}^{U}.

For a fixed incident-photon energy, two-body kinematics restricts the outgoing ALP energy to a narrow interval around Ea≃Eγ′E_a\simeq E'_\gamma. Following [10], we therefore employ the narrow-energy-transfer approximation

dσPUdEa≃σPU(Ea;U) δ(Eγ′−Ea).(28)\frac{\mathrm{d}\sigma_{\mathrm{P}}^{U}}{\mathrm{d}E_a}\simeq\sigma_{\mathrm{P}}^{U}(E_a;U)\,\delta(E'_\gamma-E_a). \tag*{(28)}

After performing the convolution with the reactor photon spectrum, the survival-weighted differential ALP rate becomes

dN˙aPdEa=Psurv(Ea)σPU(Ea;U)σSMU(Ea)+σPU(Ea;U)dN˙γdEγ′∣Eγ′=Ea.(29)\frac{\mathrm{d}\dot{N}_a^{P}}{\mathrm{d}E_a}=P_{\mathrm{surv}}(E_a)\frac{\sigma_{\mathrm{P}}^{U}(E_a;U)}{\sigma_{\mathrm{SM}}^{U}(E_a)+\sigma_{\mathrm{P}}^{U}(E_a;U)}\left.\frac{\mathrm{d}\dot{N}_\gamma}{\mathrm{d}E'_\gamma}\right|_{E'_\gamma=E_a}. \tag*{(29)}

This quantity has units of MeV−1 s−1\mathrm{MeV}^{-1}\,\mathrm{s}^{-1} and does not yet include geometrical dilution. Accordingly, the differential ALP flux incident on a detector at a distance LL is

dΦaPdEa=14πL2dN˙aPdEa.(30)\frac{\mathrm{d}\Phi_a^{P}}{\mathrm{d}E_a}=\frac{1}{4\pi L^2}\frac{\mathrm{d}\dot{N}_a^{P}}{\mathrm{d}E_a}. \tag*{(30)}

Within the single-coupling scenario considered here, the ALP decay width is dominated by the diphoton channel [8, 10],

Γa=Γa→γγ=gaγγ2ma364π,(31)\Gamma_a=\Gamma_{a\to\gamma\gamma}=\frac{g_{a\gamma\gamma}^{2}m_a^{3}}{64\pi}, \tag*{(31)}

where mam_a is the ALP mass. For an ALP with energy EaE_a and momentum pa=Ea2−ma2p_a=\sqrt{E_a^2-m_a^2}, the probability of surviving the propagation distance LL is [10, 13]

Psurv(Ea)=exp⁡[−LmaΓapaℏc],(32)P_{\mathrm{surv}}(E_a)=\exp\left[-\frac{Lm_a\Gamma_a}{p_a\hbar c}\right], \tag*{(32)}

where paℏc/(maΓa)p_a\hbar c/(m_a\Gamma_a) is the ALP decay length in the laboratory frame. The conditional probability that an ALP reaching the detector decays over an effective mean path length ℓeff\ell_{\mathrm{eff}} inside the detector (§ 5.3) is

Pdec(Ea)=1−exp⁡[−ℓeffmaΓapaℏc].(33)P_{\mathrm{dec}}(E_a)=1-\exp\left[-\frac{\ell_{\mathrm{eff}}m_a\Gamma_a}{p_a\hbar c}\right]. \tag*{(33)}

Detection mechanisms and event yields The ALP signal is the sum of two physically different contributions:

  1. Inverse Primakoff scattering (a+Ge→γ+Gea+\mathrm{Ge}\to\gamma+\mathrm{Ge}). For inverse Primakoff detection we adopt the prescription of Ref. [10],

σDP(Ea;Ge)=2 σPP(Ea;Ge).(34)\sigma_D^P(E_a;\mathrm{Ge})=2\,\sigma_P^P(E_a;\mathrm{Ge}) . \tag*{(34)}

The physical origin of the factor of two is the interchange of the photon between the initial and final states. By microscopic reversibility the same matrix element governs the direct and inverse processes; in the small-recoil kinematics Ea≃EγE_a\simeq E_\gamma used here, production averages over the two initial photon polarizations whereas the inverse process sums over the two final photon polarizations. We therefore use Eq. (34) as the reference prescription rather than as an independent exact detailed-balance calculation away from this kinematic limit. Here Eq. (17) is evaluated with ZGe=32Z_{\mathrm{Ge}}=32, AGe=72.63A_{\mathrm{Ge}}=72.63, and MGe=AGe×931.5 MeVM_{\mathrm{Ge}}=A_{\mathrm{Ge}}\times931.5\ \mathrm{MeV}. The expected number of events is

NIP=MdetNTΔt4πL2∫Elo(ma)Emax⁡dN˙aPdEa σDP(Ea;Ge) dEa,(35)N_{\mathrm{IP}}=\frac{M_{\mathrm{det}}N_T\Delta t}{4\pi L^2}\int_{E_{\mathrm{lo}}(m_a)}^{E_{\max}}\frac{\mathrm{d}\dot{N}_a^P}{\mathrm{d}E_a}\,\sigma_D^P(E_a;\mathrm{Ge})\,\mathrm{d}E_a, \tag*{(35)}

with NT=NA/(AGe×10−3 kg mol−1)N_T=N_A/(A_{\mathrm{Ge}}\times10^{-3}\ \mathrm{kg\ mol}^{-1}) the number of germanium nuclei per kilogram. Nuclear recoil is neglected in the reconstructed energy, so the outgoing photon is taken to carry essentially the full ALP energy.

  1. Decay inside the detector (a→γγa\to\gamma\gamma). An ALP intersecting the detector can instead decay into two photons. The corresponding event yield is

Nγγ=AprojΔt4πL2∫Elo(ma)Emax⁡dN˙aPdEa Pdec(Ea) dEa,(36)N_{\gamma\gamma}=\frac{A_{\mathrm{proj}}\Delta t}{4\pi L^2}\int_{E_{\mathrm{lo}}(m_a)}^{E_{\max}}\frac{\mathrm{d}\dot{N}_a^P}{\mathrm{d}E_a}\,P_{\mathrm{dec}}(E_a)\,\mathrm{d}E_a, \tag*{(36)}

where AprojA_{\mathrm{proj}} is an effective projected detector area and PdecP_{\mathrm{dec}} is evaluated with the effective mean chord ℓeff\ell_{\mathrm{eff}}, with AprojℓeffA_{\mathrm{proj}}\ell_{\mathrm{eff}} equal to the detector volume. This generalizes the A=Ldet2A=L_{\mathrm{det}}^2, LdetL_{\mathrm{det}} cube prescription of Ref. [10].

The effective geometry does not constitute a detector-response simulation; photon escape, annihilation photons, dead layers, inter-crystal gaps, passive material and event-topology selections are not modelled explicitly.

The predicted photon-coupling signal is

Saγγ=NIP+Nγγ.(37)S_{a\gamma\gamma}=N_{\mathrm{IP}}+N_{\gamma\gamma}. \tag*{(37)}

The effective geometries and the experimental parameters of CONUS+ and CONUS100 are given in § 5; the detector efficiency and background assumptions are introduced only when the projected sensitivities are constructed in § 5.5.2.

Electron–ALP coupling

We next consider the complementary single-coupling scenario in which gaeeg_{aee} is non-zero, while gaγγg_{a\gamma\gamma} and gann(0,1)g_{ann}^{(0,1)} in Eq. (12) are set to zero. Reactor photons produce ALPs through Compton-like scattering, γe−→ae−\gamma e^- \to ae^-, while detection in germanium proceeds through inverse Compton-like scattering, axio-electric absorption, or, when kinematically open, the decay a→e+e−a \to e^+e^-. These are the three channels considered here, following [10]. ALP-induced pair production in the nuclear or electron field is not included, consistently with the same reference treatment, where its expected yield is subleading.

Production and detection processes for a dominant electron–ALP coupling

Figure 5. Production and detection processes retained for a dominant electron–ALP coupling. ALPs are produced by γe−→ae−\gamma e^- \to ae^- and are detected through ae−→γe−ae^- \to\gamma e^-, axio-electric absorption, or a→e+e−a \to e^+e^- when ma>2mem_a > 2m_e.

Reactor production and propagation Our analysis uses the common reactor photon source of Eq. (1). As in the photon-coupling analysis, the production medium is represented by an effective uranium target. In the Compton-like process the relevant scatterers are the atomic electrons, so the production cross section carries the uranium atomic number ZU=92Z_U = 92. The electrons are treated as initially free and at rest; atomic-binding effects in the production medium are neglected.

For an incident reactor photon of energy Eγ′E'_\gamma, the differential cross section for γe−→ae−\gamma e^- \to ae^- is written as [10, 32]

dσPCdEa=ZUαgaee2x4(s−me2)(1−x)Eγ′[x−2ma2s(s−me2)2+2ma2(s−me2)2(me21−x+ma2x)](38)\frac{d\sigma_{P}^{C}}{dE_a} = \frac{Z_U \alpha g_{aee}^{2} x}{4(s-m_e^{2})(1-x)E'_\gamma}\left[x-\frac{2m_a^{2}s}{(s-m_e^{2})^{2}}+\frac{2m_a^{2}}{(s-m_e^{2})^{2}}\left(\frac{m_e^{2}}{1-x}+\frac{m_a^{2}}{x}\right)\right] \tag*{(38)}

where

s=me2+2meEγ′,x=1−EaEγ′+ma22meEγ′.(39)s = m_e^2 + 2m_eE'_\gamma,\qquad x = 1 - \frac{E_a}{E'_\gamma} + \frac{m_a^2}{2m_eE'_\gamma}. \tag*{(39)}

At fixed Eγ′E'_\gamma, the two-body phase space restricts xx to [33]

xmin⁡,max⁡=s−me2+ma2∓(s−me2+ma2)2−4sma22s,(40)x_{\min,\max} = \frac{s-m_e^2+m_a^2 \mp\sqrt{(s-m_e^2+m_a^2)^2-4sm_a^2}}{2s}, \tag*{(40)}

and the production threshold is

Eγ,thr′=ma+ma22me.(41)E'_{\gamma,\mathrm{thr}} = m_a + \frac{m_a^2}{2m_e}. \tag*{(41)}

The latter is particularly relevant in the MeV-mass region because the reactor spectrum decreases exponentially with Eγ′E'_\gamma and terminates at 10 MeV in the present calculation.

The photon interaction cross section in uranium is the XCOM one of Eq. (26). Since the Compton-like contribution is negligible with respect to it over the couplings considered, we use σTotU≃σSMU\sigma_{\mathrm{Tot}}^U \simeq\sigma_{\mathrm{SM}}^U in the production probability.

The survival-weighted differential ALP emission rate is then

dN˙aCdEa=Psurv(Ea)∫dEγ′ 1σSMU(Eγ′)dσPCdEadN˙γdEγ′,(42)\frac{\mathrm{d}\dot{N}_a^C}{\mathrm{d}E_a} = P_{\mathrm{surv}}(E_a) \int\mathrm{d}E'_\gamma\, \frac{1}{\sigma_{\mathrm{SM}}^U(E'_\gamma)} \frac{\mathrm{d}\sigma_P^C}{\mathrm{d}E_a} \frac{\mathrm{d}\dot{N}_\gamma}{\mathrm{d}E'_\gamma}, \tag*{(42)}

where the integration is performed only over the exact kinematically allowed incident-photon intervals implied by Eq. (40). As before, the geometrical dilution 1/(4πL2)1/(4\pi L^2) is introduced only when computing the detector event yield.

The decay a→e+e−a \to e^+e^- becomes available at ma>2mem_a > 2m_e, with tree-level width [8, 10]

Γa→e+e−=gaee2ma8π1−4me2ma2 Θ(ma−2me),(43)\Gamma_{a\to e^+e^-} = \frac{g_{aee}^2m_a}{8\pi} \sqrt{1-\frac{4m_e^2}{m_a^2}}\, \Theta(m_a-2m_e), \tag*{(43)}

where Θ\Theta denotes the Heaviside step function. Loop-induced diphoton decays are neglected in the single-gaeeg_{aee} scenario. For pa=Ea2−ma2p_a = \sqrt{E_a^2-m_a^2}, propagation from the reactor to the detector is therefore described by

Psurv(Ea)=exp⁡[−LmaΓa→e+e−paℏc],(44)P_{\mathrm{surv}}(E_a) = \exp\left[-\frac{Lm_a\Gamma_{a\to e^+e^-}}{p_a\hbar c}\right], \tag*{(44)}

while the conditional probability for an ALP reaching the detector to decay inside an effective path length ℓeff\ell_{\mathrm{eff}} is

Pdecee(Ea)=1−exp⁡[−ℓeffmaΓa→e+e−paℏc].(45)P_{\mathrm{dec}}^{ee}(E_a) = 1-\exp\left[-\frac{\ell_{\mathrm{eff}}m_a\Gamma_{a\to e^+e^-}}{p_a\hbar c}\right]. \tag*{(45)}

Below threshold, Psurv=1P_{\mathrm{surv}}=1 and Pdecee=0P_{\mathrm{dec}}^{ee}=0 within the present single-coupling approximation.

Detection mechanisms and event yields For inverse Compton-like detection, a ++ e−→γ+e−e^{-} \to\gamma+ e^{-}, we use the differential cross section for a massive ALP of Ref. [34], in the form given in Ref. [10],

dσDCdEγ=ZGegaee2αEγ4me2pa∣2(Ea+me−pacos⁡θ)2pay∣×[1+4me2Eγ2y2−4meEγy−4ma2pa2meEγ(1−cos⁡2θ)y3],(46)\frac{d\sigma^{C}_{D}}{dE_{\gamma}} = \frac{Z_{\mathrm{Ge}}g_{aee}^{2}\alpha E_{\gamma}}{4m_{e}^{2}p_{a}}\left|\frac{2(E_{a}+m_{e}-p_{a}\cos\theta)^{2}}{p_{a}y}\right|\times\left[1+\frac{4m_{e}^{2}E_{\gamma}^{2}}{y^{2}}-\frac{4m_{e}E_{\gamma}}{y}-\frac{4m_{a}^{2}p_{a}^{2}m_{e}E_{\gamma}(1-\cos^{2}\theta)}{y^{3}}\right], \tag*{(46)}

where ZGe=32Z_{\mathrm{Ge}}=32, θ\theta is the angle between the incident ALP and the outgoing photon, and

y=2meEa+ma2,cos⁡θ=Ea+me−y/(2Eγ)pa.(47)y=2m_{e}E_{a}+m_{a}^{2},\qquad\cos\theta=\frac{E_{a}+m_{e}-y/(2E_{\gamma})}{p_{a}}. \tag*{(47)}

For each EaE_{a} we integrate over the exact final-state photon interval

Eγmin⁡=max⁡[25 keV,y2(Ea+me+pa)],(48)E_{\gamma}^{\min}=\max\left[25\ \mathrm{keV},\frac{y}{2(E_{a}+m_{e}+p_{a})}\right], \tag*{(48)}
Eγmax⁡=min⁡[10 MeV,y2(Ea+me−pa)],(49)E_{\gamma}^{\max}=\min\left[10\ \mathrm{MeV},\frac{y}{2(E_{a}+m_{e}-p_{a})}\right], \tag*{(49)}

and define

σDC(Ea)=∫Eγmin⁡Eγmax⁡dσDCdEγ dEγ.(50)\sigma^{C}_{D}(E_{a})=\int_{E_{\gamma}^{\min}}^{E_{\gamma}^{\max}}\frac{d\sigma^{C}_{D}}{dE_{\gamma}}\,dE_{\gamma}. \tag*{(50)}

The second scattering channel is axio-electric absorption. Its cross section is related to the ordinary photoelectric cross section of germanium by [35, 36, 37, 38]

σDA(Ea)=gaee2β3Ea216παme2(1−β2/33)σPEGe(Ea),β=paEa,(51)\sigma^{A}_{D}(E_{a})=\frac{g_{aee}^{2}}{\beta}\frac{3E_{a}^{2}}{16\pi\alpha m_{e}^{2}}\left(1-\frac{\beta^{2/3}}{3}\right)\sigma^{\mathrm{Ge}}_{\mathrm{PE}}(E_{a}),\qquad\beta=\frac{p_{a}}{E_{a}}, \tag*{(51)}

where σPEGe\sigma^{\mathrm{Ge}}_{\mathrm{PE}} is taken from the germanium XCOM table. The corresponding scattering event yields are

NIC=MdetNTΔt4πL2∫dEadN˙aCdEaσDC(Ea),(52)N_{\mathrm{IC}}=\frac{M_{\mathrm{det}}N_{T}\Delta t}{4\pi L^{2}}\int dE_{a}\frac{d\dot{N}^{C}_{a}}{dE_{a}}\sigma^{C}_{D}(E_{a}), \tag*{(52)}
NAE=MdetNTΔt4πL2∫dEadN˙aCdEaσDA(Ea),(53)N_{\mathrm{AE}}=\frac{M_{\mathrm{det}}N_{T}\Delta t}{4\pi L^{2}}\int dE_{a}\frac{d\dot{N}^{C}_{a}}{dE_{a}}\sigma^{A}_{D}(E_{a}), \tag*{(53)}

where NTN_{T} is the number of germanium atoms per unit detector mass, as defined in § 4.2.1.

For ma>2mem_{a}>2m_{e}, ALPs can also be detected through an in-detector electron–positron decay. We use the same effective decay geometry, characterized by AprojA_{\mathrm{proj}} and ℓeff\ell_{\mathrm{eff}}, introduced in § 4.2.1. The corresponding yield is

Nee=AprojΔt4πL2∫dEadN˙aCdEaPdecee(Ea).(54)N_{ee}=\frac{A_{\mathrm{proj}}\Delta t}{4\pi L^{2}}\int dE_{a}\frac{d\dot{N}^{C}_{a}}{dE_{a}}P^{ee}_{\mathrm{dec}}(E_{a}). \tag*{(54)}

The total single-coupling electron–ALP signal is therefore

Saee=NIC+NAE+Nee.(55)S_{aee}=N_{\mathrm{IC}}+N_{\mathrm{AE}}+N_{ee}. \tag*{(55)}

Nucleon–ALP coupling

We finally consider ALP production through the nucleon interactions already introduced in Eq. (12). In contrast to Primakoff and Compton-like production, this channel is generated by discrete magnetic nuclear transitions rather than by the continuous reactor photon spectrum of Eq. (1). Following neutron capture or the production of excited fission fragments, an excited nuclear state N∗N^{*} may emit an ALP instead of a photon. Pseudoscalar emission competes with magnetic MJM J transitions, whereas electric transitions do not contribute through the nucleon coupling [10, 14, 15]. Figure 6 shows the production process N∗→N+aN^{*} \to N + a and its formal inverse; the latter is displayed only for completeness and is not used as a detection channel here. The magnetic transitions and their photon yields are described in § 2.2.

Nuclear de-excitation and ALP productionFormal inverse process

Figure 6. Nuclear processes associated with the nucleon–ALP coupling. Panel (a) shows the production mechanism used in this work, N∗→N+aN^{*} \to N + a. Panel (b) shows the formal inverse reaction a+N→N∗→N+γa + N \to N^{*} \to N + \gamma, which is not used for detection because the germanium target does not provide the resonant isotopes and transitions required to absorb the reactor ALP lines.

ALP-to-photon branching ratios For a transition ii, ALP production is determined by the ratio of the ALP and photon partial widths, Γia/Γiγ\Gamma_i^a/\Gamma_i^\gamma. We distinguish these nuclear-transition widths from the total particle decay width Γatot\Gamma_a^{\mathrm{tot}}, which controls propagation. For the isovector M1M1 transition p(n,γ)dp(n,\gamma)d, the ratio is [10, 39]

(ΓiaΓiγ)pn=12πα(papγ)3(gann(1)μ1)2(56)\left(\frac{\Gamma_i^a}{\Gamma_i^\gamma}\right)_{pn} = \frac{1}{2\pi\alpha}\left(\frac{p_a}{p_\gamma}\right)^3\left(\frac{g_{ann}^{(1)}}{\mu_1}\right)^2 \tag*{(56)}

where μ1=μp−μn=4.71\mu_1=\mu_p-\mu_n=4.71 is the isovector nuclear magnetic moment in units of the nuclear magneton [36]. Neglecting the very small nuclear recoil, pγ≃Eip_\gamma\simeq E_i and pa=Ei2−ma2p_a=\sqrt{E_i^2-m_a^2}. Therefore, emission from line ii is allowed only for ma<Eim_a<E_i.

For a general magnetic transition of multipolarity MJM J, the long-wavelength result is [10, 17, 36]

(ΓiaΓiγ)MJ=1πα11+δi2JJ+1(papγ)2J+1[gann(0)κi+gann(1)(μ0−12)κi+(μ1+ηi)]2(57)\left(\frac{\Gamma_i^a}{\Gamma_i^\gamma}\right)_{M J} = \frac{1}{\pi\alpha}\frac{1}{1+\delta_i^2}\frac{J}{J+1}\left(\frac{p_a}{p_\gamma}\right)^{2J+1}\left[\frac{g_{ann}^{(0)}\kappa_i+g_{ann}^{(1)}}{\left(\mu_0-\frac{1}{2}\right)\kappa_i+\left(\mu_1+\eta_i\right)}\right]^2 \tag*{(57)}

where JJ is the magnetic multipole order and μ0=μp+μn=0.88\mu_{0} = \mu_{p} + \mu_{n} = 0.88 is the isoscalar nuclear magnetic moment, also in nuclear-magneton units. The parameter δi\delta_{i} is the electric-to-magnetic multipole mixing ratio, while ηi\eta_{i} and κi\kappa_{i} encode ratios of nuclear spin and orbital matrix elements [36]. For the predominantly M1 7Li∗{}^{7}\mathrm{Li}^{*} transition, we use δi≪1\delta_{i} \ll1, ηi=0.5\eta_{i} = 0.5, and κi=1.0\kappa_{i} = 1.0, as calculated in Ref. [21] and adopted by TEXONO [17]. For the M4 lines no calculations are available; TEXONO estimated them from the results for 65Cu{}^{65}\mathrm{Cu}, 57Fe{}^{57}\mathrm{Fe}, 55Mn{}^{55}\mathrm{Mn} and 23Na{}^{23}\mathrm{Na} of Refs. [36, 40], obtaining δi=0\delta_{i} = 0, ηi=1\eta_{i} = 1, and κi=−3\kappa_{i} = -3 for 91Y∗{}^{91}\mathrm{Y}^{*} and 97Nb∗{}^{97}\mathrm{Nb}^{*}, and δi=0\delta_{i} = 0, ηi=−1\eta_{i} = -1, and κi=1\kappa_{i} = 1 for 135Xe∗{}^{135}\mathrm{Xe}^{*} and 137Ba∗{}^{137}\mathrm{Ba}^{*} [17]. We adopt the same values as Ref. [10]. These nuclear-structure inputs are adopted as fixed benchmark values, and their uncertainties are not propagated as nuisance parameters.

Monochromatic ALP source and propagation The transition energy EiE_{i} is fixed by the nuclear level spacing. After neglecting nuclear recoil and the intrinsic line width, the corresponding ALP source is treated as monochromatic. The photon yields Φγ,i\Phi_{\gamma,i} and the fission rate Rf(P)R_{f}(P) are those of § 2.2, Eq. (2).

Defining

Bia≡ΓiaΓiγ,(58)\mathcal{B}_{i}^{a} \equiv\frac{\Gamma_{i}^{a}}{\Gamma_{i}^{\gamma}}, \tag*{(58)}

the survival-weighted differential ALP emission rate from transition ii is

dN˙a,iMTdEa=Rf(P) Φγ,iBiaPsurv(Ei)δ(Ea−Ei)Θ(Ei−ma).(59)\frac{\mathrm{d}\dot{N}_{a,i}^{\mathrm{MT}}}{\mathrm{d}E_{a}} = R_{f}(P)\,\Phi_{\gamma,i}\mathcal{B}_{i}^{a}P_{\mathrm{surv}}(E_{i})\delta(E_{a}-E_{i})\Theta(E_{i}-m_{a}). \tag*{(59)}

Here the superscript “MT” denotes production through a magnetic nuclear transition. Equation (59) uses the weak-branching approximation Γia≪Γiγ\Gamma_{i}^{a} \ll\Gamma_{i}^{\gamma}, which is valid throughout the parameter region considered. In the mixed-coupling projections below, propagation is evaluated with the decay width associated with the detection coupling: Γa→γγ\Gamma_{a\to\gamma\gamma} for photon-mediated detection and Γa→e+e−\Gamma_{a\to e^{+}e^{-}} for electron-mediated detection. The corresponding survival probabilities are those defined in Eqs. (32) and (44), respectively. In the long-lived region both approach unity.

Detection in germanium Detection through resonant nuclear excitation would require a target containing the appropriate isotopes and matching transitions. Since the germanium detectors do not satisfy this requirement for the six reactor lines considered here, the nuclear-production channel cannot constrain gann(0,1)g_{ann}^{(0,1)} by itself. Instead, the produced ALPs are detected through a second coupling: inverse Primakoff conversion for gaγγg_{a\gamma\gamma}, or inverse Compton-like scattering and axio-electric absorption for gaeeg_{aee}. In-detector decays into photons or electron–positron pairs can also contribute when they are kinematically allowed. The resulting analysis therefore probes coupling pairs rather than an isolated nucleon coupling.

For compactness, define the survival-weighted line rate

φa,i=Rf(P) Φγ,iBiaPsurv(Ei)Θ(Ei−ma),(60)\varphi_{a,i} = R_{f}(P)\,\Phi_{\gamma,i}\mathcal{B}_{i}^{a}P_{\mathrm{surv}}(E_{i})\Theta(E_{i}-m_{a}), \tag*{(60)}

which has units of s−1\mathrm{s}^{-1}. Using the detection cross sections defined in the preceding subsections, the expected number of events from line ii is

NP,iMT=MdetNTΔt4πL2ϕa,iσDP(Ei),(61)N^{\mathrm{MT}}_{P,i} = \frac{M_{\mathrm{det}}N_T\Delta t}{4\pi L^2}\phi_{a,i}\sigma_D^P(E_i), \tag*{(61)}
NC,iMT=MdetNTΔt4πL2ϕa,iσDC(Ei),(62)N^{\mathrm{MT}}_{C,i} = \frac{M_{\mathrm{det}}N_T\Delta t}{4\pi L^2}\phi_{a,i}\sigma_D^C(E_i), \tag*{(62)}
NA,iMT=MdetNTΔt4πL2ϕa,iσDA(Ei),(63)N^{\mathrm{MT}}_{A,i} = \frac{M_{\mathrm{det}}N_T\Delta t}{4\pi L^2}\phi_{a,i}\sigma_D^A(E_i), \tag*{(63)}

where the inverse-Compton cross section σDC(Ei)\sigma_D^C(E_i) is understood to be integrated over the accepted final-state energy interval.

For a visible decay mode ff, the corresponding yield is

Ndec,iMT,f=AprojΔt4πL2ϕa,iPdecf(Ei),(64)N^{\mathrm{MT},f}_{\mathrm{dec},i} = \frac{A_{\mathrm{proj}}\Delta t}{4\pi L^2}\phi_{a,i}P_{\mathrm{dec}}^f(E_i), \tag*{(64)}

with

Pdecf(Ei)=Γa→fΓatot[1−exp⁡(−ℓeffmaΓatotpaℏc)].(65)P_{\mathrm{dec}}^f(E_i) = \frac{\Gamma_{a\to f}}{\Gamma_a^{\mathrm{tot}}}\left[1-\exp\left(-\frac{\ell_{\mathrm{eff}}m_a\Gamma_a^{\mathrm{tot}}}{p_a\hbar c}\right)\right]. \tag*{(65)}

Here pa=Ei2−ma2p_a=\sqrt{E_i^2-m_a^2} and Γatot\Gamma_a^{\mathrm{tot}} denotes the total ALP decay width relevant for the mixed-coupling scenario. The decay term uses the same AprojA_{\mathrm{proj}} and ℓeff\ell_{\mathrm{eff}} geometries adopted in the single-coupling photon and electron analyses. If only one decay mode is open, the branching-fraction prefactor in Eq. (65) equals unity.

The six transitions above are retained to characterize the magnetic-line reactor source. For the mixed-coupling sensitivity projections in § 5.5.2, however, we follow [10] and retain only the dominant isovector p(n,γ)dp(n,\gamma)d contribution. We furthermore adopt

gann(0)=0,(66)g_{ann}^{(0)}=0, \tag*{(66)}

so that nuclear production is controlled by gann(1)g_{ann}^{(1)}. In the weak-coupling, long-lived limit, inverse-Primakoff detection then scales as (gann(1)gaγγ)2(g_{ann}^{(1)}g_{a\gamma\gamma})^2, while inverse-Compton-like and axio-electric detection scale as (gann(1)gaee)2(g_{ann}^{(1)}g_{aee})^2. Away from this limit, the propagation and decay probabilities retain a separate dependence on the detection coupling.

Application to CONUS+ and CONUS100

The framework of § 2–§ 4 is now applied to the CONUS+ experiment and to the proposed CONUS100 upgrade.

Experimental setup

CONUS+ was installed in 2023 at the Leibstadt nuclear power plant (KKL), Switzerland, as the successor of the CONUS experiment at the Brokdorf reactor (KBR), Germany. Its primary goal is the study of coherent elastic neutrino–nucleus scattering (CEν\nuNS) and the search for deviations from the SM prediction, with the boiling-water reactor as antineutrino source. The same environment is favourable for dark photon and ALP searches: the reactor operates at a maximum thermal power of 3.6 GW, and the antineutrino flux at the detector site, 1.45×1013 cm−2 s−11.45 \times10^{13}\,\mathrm{cm}^{-2}\,\mathrm{s}^{-1}, reflects the high fission rate (Figure 7) and hence the intense photon emission of the core.

Exemplary average fission rate fractions of the four dominant fissile actinides

Figure 7. Exemplary average fission rate fractions of the four dominant fissile actinides as function of one typical reactor cycle evolution at KKL. Retrieved from [19].

In the CONUS+ area, the average radiation dose is below 1 μSv h−11\,\mu\mathrm{Sv}\,\mathrm{h}^{-1} and the average overburden is 7.4 m of water equivalent [19].

The detector setup is situated at a distance of ∼20.7\sim20.7 m from the reactor core. It consists of 4 High Purity Germanium (HPGe) detectors C2, C3, C4, C5, each having 1 kg mass and resulting in an active mass of 0.92–0.95 kg per detector. The decreasing trigger efficiency towards very low energies is substantially compensated by the small point contact size design choice of the P-type Point Contact (PPC) detectors. This not only decreases the capacitance, but also features a highly steep field gradient near the point contact, improving further the pulse shape discrimination. Furthermore, in order to overcome the undesired effects of thermal excitation to a good extent, the detectors are kept inside Copper cryostats connected to cryocoolers outside the shield. The high-purity crystal choice is also crucial for the minimization of the leakage current [41].

Outside the detectors, an onion-like structure of shield layers as in Figure 8 is present for mitigation of several background components:

CONUS+ setup

Figure 8. CONUS+ setup: Adapted from [19].

  • (i) Four layers of Pb as passive shield,

  • (ii) Two plates of polyethylene and two layers of B-doped polyethylene as neutron shield, and

  • (iii) Two layers of plastic scintillator plates as active μ\mu veto

This whole setup is housed inside a steel frame accounting for the strict earthquake requirements of the site. Radon diffusion is reduced by the radon-seal tape at the frame edges.

Detector Performance

The resolution and trigger efficiency performances of each CONUS+ detector are given respectively in Table 1 and Figure 9.

DetectorPulser resolution
C248±1 eVee48 \pm1\,eV_{ee}
C347±1 eVee47 \pm1\,eV_{ee}
C447±1 eVee47 \pm1\,eV_{ee}
C547±1 eVee47 \pm1\,eV_{ee}

Table 1. Pulser resolution measurement results of each CONUS+ detector at the KKL site. Adapted from [19].

Trigger efficiency vs. energy curves for each CONUS+ detector

Figure 9. Trigger efficiency vs. energy curves for each CONUS+ detector. Adapted from [19].

The expected CEνNSCE\nu NS signal as a function of the energy threshold for the different CONUS and CONUS+ phases is shown in Figure 10.

CEνNS signal prediction vs. energy threshold for CONUS Run-5, CONUS+ Phase 1, and CONUS+ Phase 2

Figure 10. CEνNSCE\nu NS signal prediction vs. energy threshold for CONUS Run-5, CONUS+ Phase 1, and CONUS+ Phase 2. Adapted from [19].

CONUS100 is a proposed upgrade of the CONUS+ technology to a 100 kg germanium array [20]. As benchmark we take P=3.5P = 3.5 GW, L=20L = 20 m and Mdet=100M_{\mathrm{det}} = 100 kg. For the dark photon and ALP projections its readout is assumed to extend to 10 MeV in deposited energy, while CONUS+ keeps its present 0.6 MeV DAQ range. The projected CEνNSCE\nu NS threshold of 120 eV is not relevant here, since the signals considered lie in the keV–MeV range.

For the ALP projections based on § 4, we use the common experimental benchmarks summarized in Table 2. The quoted exposures are full-power-equivalent (FPE) exposures; reactor outages, dead time, and detector efficiencies are treated separately as analysis assumptions rather than as experimental parameters. In particular, the live-time fraction of CONUS+, 0.88 (muon-veto dead time), is not applied in the dark photon and ALP projections. For CONUS+ we keep the present 0.6 MeV DAQ range, i.e. unit efficiency between 25 keV and 0.6 MeV, since the published readout (channels up to 35 keVee_{\mathrm{ee}} and 600 keVee_{\mathrm{ee}}) is optimised for CEν\nuNS [19]. For CONUS100 the ALP projections instead assume that the readout is extended to 10 MeV, with unit efficiency between 25 keV and 10 MeV, since the ALP signals extend to MeV energies. The same windows are used for the dark photon search, with a lower edge of 0.2 MeV (§ 5.5.1 and § 5.5.2). Including the live-time fraction would weaken the projected couplings by only 1–3%. With Eq. (2), the fission rates are Rf=1.1232×1020 s−1R_f = 1.1232 \times10^{20}\ \mathrm{s}^{-1} for KKL (P=3.6P = 3.6 GW) and 1.092×1020 s−11.092 \times10^{20}\ \mathrm{s}^{-1} for the CONUS100 benchmark (P=3.5P = 3.5 GW).

Effective detector geometry

The decay yields of Eqs. (36), (54) and (64) require an effective projected area AprojA_{\mathrm{proj}} and mean chord ℓeff\ell_{\mathrm{eff}} of the germanium target. For CONUS+, the four refurbished HPGe detectors are modeled as four cylindrical crystals of height and diameter 62 mm [19]. Since the orientation of the crystals relative to the reactor is not modelled, we use the orientation-averaged projected area of a convex body, ⟨Aproj⟩=S/4\langle A_{\mathrm{proj}}\rangle= S/4, where SS is the crystal surface area, and sum the four crystals without overlap. This gives

AprojC+=181.14 cm2,ℓeffC+=4.13 cm,(67)A_{\mathrm{proj}}^{\mathrm{C+}} = 181.14\ \mathrm{cm}^{2}, \qquad\ell_{\mathrm{eff}}^{\mathrm{C+}} = 4.13\ \mathrm{cm}, \tag*{(67)}

with AprojℓeffA_{\mathrm{proj}}\ell_{\mathrm{eff}} equal to the modeled Ge volume. The inverse-Primakoff normalization uses the measured active mass, whereas the decay term uses the physical crystal dimensions as an effective geometrical interception proxy.

For CONUS100 no final crystal layout is presently specified in the proposal [20]. We therefore use a compact cube having the volume of 100 kg of Ge. With the standard Ge density this gives a side length 26.58 cm and hence

AprojC100=706.69 cm2,ℓeffC100=26.58 cm.(68)A_{\mathrm{proj}}^{\mathrm{C100}} = 706.69\ \mathrm{cm}^{2}, \qquad\ell_{\mathrm{eff}}^{\mathrm{C100}} = 26.58\ \mathrm{cm}. \tag*{(68)}

This CONUS100 geometry is a projection benchmark, not an official detector dimension. For the cube we use the face-on projection, whereas for CONUS+ we use the orientation average; this difference has no effect on the results. An ALP that reaches the detector has a decay length much larger than ℓeff\ell_{\mathrm{eff}}, since otherwise it would have decayed along the 20 m baseline, so that Pdec≃ℓeff/λdecP_{\mathrm{dec}} \simeq\ell_{\mathrm{eff}}/\lambda_{\mathrm{dec}} and the decay yield depends only on the product AprojℓeffA_{\mathrm{proj}}\ell_{\mathrm{eff}}, i.e. on the detector volume.

The modeled CONUS+ volume, 749 cm3^{3}, corresponds to 3.99 kg of Ge, i.e. 6.6% above the active mass used for the scattering yields; restricting the decays to the active volume would lower the decay yields by this factor and weaken the decay-dominated reach by about 3%. Note also that the cube of Ref. [10] (Ldet=50L_{\mathrm{det}} = 50 cm for 10 kg of Ge) has a volume much larger than that of the detector, so the decay rates of that reference are not directly comparable with the ones derived here.

Expected event rates

Figure 11, Figure 12 and Figure 13 show representative differential event rates in CONUS+ and CONUS100 for dark photons and for ALPs coupled to photons and to electrons, respectively. The dark photon rates use the same production and detection cross sections as the sensitivity of § 5.5.1, with the CONUS+ and CONUS100 parameters of Table 2; they scale as ϵ4\epsilon^{4}. The ALP rates are shown before any energy window is applied: for CONUS+ only the part below 0.6 MeV enters the projections, so that the a→e+e−a \to e^{+}e^{-} signal lies entirely outside its window, whereas for CONUS100 the window is assumed to extend to 10 MeV. The small structure near 0.1 MeV in the photon-coupling rates reflects the uranium K edge in the photon attenuation of the production medium. The sharp edges of the electron-coupling rates are kinematic limits of the production and detection processes; for instance, the inverse Compton-like photon spectrum is bounded from below by ≃me/2\simeq m_{e}/2, and for ma>mem_{a} > m_{e} the Compton-like production gives Ea≳(ma2+me2)/(2me)E_{a} \gtrsim(m_{a}^{2} + m_{e}^{2})/(2m_{e}).

Differential event rates for inverse Primakoff scattering and decays

Figure 12. Differential event rates per keV and per full-power year for ma=10 keVm_a = 10\ \mathrm{keV} and gaγγ=10−3 GeV−1g_{a\gamma\gamma} = 10^{-3}\ \mathrm{GeV}^{-1}: inverse Primakoff scattering (left) and a→γγa \to\gamma\gamma decays inside the detector (right), for CONUS+ (solid) and CONUS100 (dashed). No efficiency or energy window is applied; the dotted line marks the upper end of the CONUS+ DAQ window, 0.6 MeV.

Differential event rates for CONUS+ and CONUS100

Figure 13. Differential event rates per keV and per full-power year for CONUS+ (left) and CONUS100 (right): inverse Compton-like scattering (top) and axio-electric absorption (middle) for gaee=10−4g_{aee}=10^{-4}, and a→e+e−a \to e^{+}e^{-} decays inside the detector (bottom) for gaee=10−6g_{aee}=10^{-6}; the line style gives mam_a. No efficiency or energy window is applied; the shaded area marks the energies above the CONUS+ DAQ window, 0.6 MeV.

Projected sensitivities

The sensitivities presented below are obtained from the signal predictions derived in the previous sections. Different confidence-level conventions are retained for the two scenarios considered in this work: the dark photon projection is quoted at 95% C.L., following the convention adopted in Ref. [7], whereas the ALP projections are presented at 90% C.L., consistently with Ref. [10].

Dark photon

The projected sensitivity to the kinetic-mixing parameter ϵ\epsilon is obtained from the signal prediction of Eq. (10). We assume a flat benchmark background of b=10b = 10 DRU (1 DRU = 1 event kg−1^{-1} day−1^{-1} keV−1^{-1}) within the accepted deposited energy interval, 0.2–0.6 MeV for CONUS+ (present DAQ) and 0.2–10 MeV for CONUS100 (assumed extended readout), with the same upper edges as for the ALPs; the lower edge is the one adopted for the photon spectrum in Ref. [7]. The expected number of background events is therefore B=bMdetΔtΔEB = b M_{\mathrm{det}} \Delta t \Delta E, where MdetM_{\mathrm{det}} is the active detector mass, Δt\Delta t the exposure time and ΔE\Delta E the width of the accepted interval.

Following the counting prescription of Ref. [7], the expected 95% C.L. upper limit on a background-only signal is approximated by

S95=1.64B,(69)S_{95} = 1.64\sqrt{B}, \tag*{(69)}

the one-sided 95% quantile appropriate for an upper limit; Ref. [7] used the two-sided value 1.96, as noted in Ref. [22]. Using the ϵ4\epsilon^4 dependence of the total event yield, Eq. (11), the corresponding sensitivity is

ϵ95(mA′)=[S95N0(mA′)]1/4,(70)\epsilon_{95}(m_{A'}) = \left[\frac{S_{95}}{N_0(m_{A'})}\right]^{1/4}, \tag*{(70)}

with N0(mA′)N_0(m_{A'}) from Eq. (11).

The resulting sensitivities are shown in Figure 14 over the interval 100 eV≤mA′≤2me100\ \mathrm{eV} \leq m_{A'} \leq2m_e. At low masses N0N_0 depends only weakly on mA′m_{A'}, and the sensitivity forms a plateau at ϵ≃1.65×10−5\epsilon\simeq1.65 \times10^{-5} for CONUS+ and 1.2×10−51.2 \times10^{-5} for CONUS100. At mA′∼0.1m_{A'} \sim0.1–0.3 MeV0.3\ \mathrm{MeV} the finite-mass kinematics redistributes the dark photon spectrum within the accepted window, and the sensitivity improves slightly, to 1.45×10−51.45 \times10^{-5} and 1.08×10−51.08 \times10^{-5} at mA′≃0.3 MeVm_{A'} \simeq0.3\ \mathrm{MeV}.

Projected sensitivity limits for CONUS+ and CONUS100

Figure 14. Projected 95% C.L. sensitivity to ϵ\epsilon for CONUS+ (solid) and CONUS100 (dashed), with b=10b = 10 DRU and unit efficiency between 0.2 and 0.6 MeV (CONUS+, present DAQ) or 10 MeV (CONUS100, assumed). Shaded regions are laboratory limits: TEXONO as re-analysed in Ref. [22] (green), atomic spectroscopy [42] (aqua) and solar dark photons in XENON1T [43] (orange), from Ref. [44].

For mA′>Edepmin⁡m_{A'} > E_{\mathrm{dep}}^{\min}, the available integration interval progressively decreases because its lower boundary rises with mA′m_{A'}. For CONUS+ the corresponding loss of accepted phase space eventually dominates the mass dependence of the production and detection rates, causing the sensitivity to deteriorate rapidly. In the limit mA′→Edepmax⁡=0.6 MeVm_{A'} \to E_{\mathrm{dep}}^{\max} = 0.6\ \mathrm{MeV} the accepted interval collapses, N0(mA′)→0N_0(m_{A'}) \to0, and the projected upper limit on ϵ\epsilon diverges, so the CONUS+ curve terminates at this endpoint. With its readout extended to 10 MeV, CONUS100 keeps its sensitivity up to the A′→e+e−A' \to e^{+}e^{-} threshold, reaching ϵ≃2.3×10−5\epsilon\simeq2.3 \times10^{-5} at mA′=1 MeVm_{A'} = 1\ \mathrm{MeV}.

The relative improvement obtained with CONUS100 can also be understood from the scaling of the signal and background. In the background-dominated regime,

ϵ95∝(MdetΔt)−1/8(L2P)1/4,(71)\epsilon_{95} \propto(M_{\mathrm{det}}\Delta t)^{-1/8}\left(\frac{L^2}{P}\right)^{1/4}, \tag*{(71)}

for fixed background density and accepted energy interval. The larger target mass and exposure of the CONUS100 benchmark alone would lead to an improvement of approximately a factor of 1.9 in the low-mass sensitivity relative to CONUS+. Since most of the signal lies below 1 MeV while the background grows with the width of the window, ϵ95∝(ΔE)1/8\epsilon_{95} \propto(\Delta E)^{1/8}, the wider CONUS100 window reduces this factor to 1.4 at low mass, the

price of its larger mass reach. A dedicated region of interest of 0.2–1.0 MeV would give ϵ≃(0.8−0.9)×10−5\epsilon\simeq(0.8-0.9) \times10^{-5} for mA′≲0.5m_{A^\prime} \lesssim0.5 MeV.

For comparison, Figure 14 displays the laboratory limits in this mass range [44]: the TEXONO reactor search as re-analysed by Danilov et al. [22], ϵ<2.6×10−5\epsilon< 2.6 \times10^{-5} for mA′≳20m_{A^\prime} \gtrsim20 eV, and atomic spectroscopy [42], which reaches 1.7×10−51.7 \times10^{-5} near 5 keV. As for the ALPs, we also show the search for solar dark photons absorbed in XENON1T [43], which excludes ϵ≳10−14\epsilon\gtrsim10^{-14}–10−1210^{-12} up to mA′≃0.5m_{A^\prime} \simeq0.5 keV, the largest mass for which the solar flux was computed there. Searches that assume that the dark photon is the dark matter are not shown. Including the secondary photons of Ref. [24] would strengthen the TEXONO limit by at most ∼30%\sim30\% [25]. Stellar and other astrophysical exclusions are not included. For 0.5 keV ≲mA′≲0.5\lesssim m_{A^\prime} \lesssim0.5 MeV, CONUS+ lies below the TEXONO limit by a factor ≃1.6\simeq1.6 and CONUS100 by ≃2.2\simeq2.2 in ϵ\epsilon. As its window closes, CONUS+ crosses the TEXONO limit at mA′≃0.56m_{A^\prime} \simeq0.56 MeV, whereas CONUS100 stays below it up to 2me2m_e; near 5 keV the CONUS+ projection touches the spectroscopy bound.

Axion-like particles

The ALP signal yields derived in Sec. 4 are translated into projected sensitivities with the same flat background benchmark as for the dark photon, b=10b = 10 DRU, for all detection channels. This value is a phenomenological benchmark and should not be interpreted as a measurement of the CONUS+ background over the full energy range relevant to the different channels. For a continuum signal the expected background is BX=bMdetΔtΔEXB_{\mathrm{X}} = b M_{\mathrm{det}} \Delta t \Delta E_{\mathrm{X}}, where ΔEX\Delta E_{\mathrm{X}} is the full accepted energy interval of the corresponding signal; Ref. [10] does not specify the interval used for the background. For the monochromatic signatures of the magnetic-transition analysis (inverse Primakoff conversion, axio-electric absorption and in-detector decays of the 2.230 MeV ALPs), the DRU density is converted to a single-bin background using a 1 keV analysis bin. This choice is a convention of the present projection and should not be interpreted as a detector resolution or as an experimentally optimized region of interest. For a germanium resolution of about 2 keV (FWHM) at 2.23 MeV, a bin containing essentially the whole line would be ≃3\simeq3 keV wide, which would weaken the projected reach on the coupling products by about 30%. The escape of annihilation photons after a→e+e−a \to e^{+}e^{-} decays, which would move part of the signal out of the full-energy line, is not modelled.

Following Ref. [10], the projected 90% C.L. sensitivity is determined from

χ2=S2S+B,Δχ2=2.71,(72)\chi^{2} = \frac{S^{2}}{S+B}, \qquad\Delta\chi^{2} = 2.71, \tag*{(72)}

where SS and BB denote the expected signal and background event counts, respectively. Throughout the ALP projections we assume unit signal efficiency inside the accepted energy window of Table 2: 25 keV–0.6 MeV for CONUS+, set by its present DAQ [19], and 25 keV–10 MeV for CONUS100, which assumes an extended readout, and the full-power-equivalent exposures of Table 2. For the ALP masses and couplings considered here the ALP energy is fully deposited in the germanium, Edep=EaE_{\mathrm{dep}} = E_{a}, so for CONUS+ only ALPs with Ea≤0.6E_{a} \leq0.6 MeV contribute, and the continuum background BXB_{\mathrm{X}} is integrated only up to 0.6 MeV. Two signals are therefore outside the CONUS+ window: the a→e+e−a \to e^{+}e^{-} decays, which deposit at least 2me≃1.0222m_{e} \simeq1.022 MeV, and all signals from the 2.230 MeV p(n,γ)dp(n,\gamma)d line. CONUS+ thus appears only in the single-coupling projections of Figure 15 and Figure 16, and the nuclear-transition and a→e+e−a \to e^{+}e^{-} results are shown for CONUS100 alone. Systematic uncertainties and nuisance parameters are not included, so the resulting curves should be interpreted as idealized sensitivity projections rather than detector-level likelihood results.

Projected sensitivity to $|g_{a\gamma\gamma}|$ for CONUS+ and CONUS100

Figure 15. Projected 90% C.L. sensitivity to ∣gaγγ∣|g_{a\gamma\gamma}| for CONUS+ (solid) and CONUS100 (dashed), with b=10b = 10 DRU and unit efficiency between 25 keV and 0.6 MeV (CONUS+, present DAQ) or 10 MeV (CONUS100, assumed). Shaded regions are experimental limits: CAST [52, 53] (yellow), BaBar [46] (red), the beam dumps E137 [46, 47], E141 [48], NuCal [49], CHARM [50] and NA64 [51], as combined in Ref. [44] (pink), and NEON [18] (blue, 95% C.L.).

Projected sensitivity to $|g_{aee}|$ for CONUS+ and CONUS100

Figure 16. Projected 90% C.L. sensitivity to ∣gaee∣|g_{aee}| for CONUS+ (solid) and CONUS100 (dashed), with b=10b = 10 DRU and efficiency =1= 1 between 25 keV and 0.6 MeV (CONUS+, present DAQ) or 10 MeV (CONUS100, assumed). Shaded regions are experimental limits: solar-ALP searches with EDELWEISS-III [57] and XENONnT [56] (khaki); the electron beam dumps E137 [47], E141 [48] and Bechis et al. [59] (pink, one region as in Ref. [18]); E774 [60] (orange) and Orsay [61] (teal), from Ref. [62]; NA64 missing-energy (purple) and visible-decay (green) searches, from Ref. [58]; and NEON [18] (blue, 95% C.L.).

In Figure 15–Figure 19 the projections are compared only with current experimental limits, i.e. searches with a laboratory source (reactors, accelerators, beam dumps) or with a terrestrial detector exposed to the solar ALP flux (helioscopes, Borexino, BGO, SNO and 57Fe{}^{57}\mathrm{Fe} searches). Bounds from stellar cooling, supernovae and cosmology, as well as searches that assume that the ALP constitutes the galactic dark matter, are not shown. All limits are at 90% C.L., except the NEON reactor limits [18] and the SNO bound [45], which are quoted at 95% C.L. In all figures, solid (dashed) curves correspond to CONUS+ (CONUS100), as indicated in a box inside each plot; when several detection channels are shown, their colours are given in the legend below the plot.

Projected sensitivities of CONUS100 and experimental limits

Figure 19. Projected 90% C.L. sensitivities of CONUS100 (dashed) in the (gaγγ,gann(1))(g_{a\gamma\gamma}, g_{a\mathrm{nn}}^{(1)}) (top) and (gaee,gann)(g_{aee}, g_{a\mathrm{nn}}) (bottom) planes at ma=1.2m_a = 1.2 MeV (left) and 10 keV (right), with b=10b = 10 DRU in a 1 keV bin and efficiency assumed to be 1 up to 10 MeV (assumed). CONUS+ is not shown, since the 2.230 MeV line lies outside its 0.6 MeV DAQ window. Shaded regions are experimental limits: SNO [45] (grey); on gaγγg_{a\gamma\gamma}, BaBar [46] (red), the beam dumps E137, E141, NuCal, CHARM and NA64 [44, 47, 48, 49, 50, 51] (pink) and NEON [18] (blue); on gaeeg_{aee}, the electron beam dumps E137, E141 and Bechis et al. [47, 48, 58, 59] (pink), NEON (blue) and the NA64 missing-energy (purple) and visible-decay (green) searches [18, 58]; and the product bounds from TEXONO [17] (light green), Borexino [63] (magenta; hatched: a→γγa \to\gamma\gamma decays of solar ALPs), BGO [64] (lavender) and the 57Fe{}^{57}\mathrm{Fe} searches [57, 65, 66, 67] (brown).

Photon–ALP coupling. Figure 15 shows the projected sensitivity to gaγγg_{a\gamma\gamma} when the same coupling controls both reactor production and detection. At low ALP masses, both CONUS+ and CONUS100 exhibit a mass-independent sensitivity, ∣gaγγ∣≃9.6×10−4|g_{a\gamma\gamma}| \simeq9.6 \times10^{-4} (3.5×10−43.5 \times10^{-4}) GeV−1^{-1} for CONUS+ (CONUS100). This corresponds to the long-lived regime, where Primakoff production followed by inverse Primakoff scattering dominates and varies only weakly with mam_a. For ma≳50m_a \gtrsim50 keV the a→γγa \to\gamma\gamma decay inside the germanium becomes the dominant channel and the sensitivity improves approximately as ma−2m_a^{-2}, reaching 1.5×10−51.5 \times10^{-5} GeV−1^{-1} at ma≃2m_a \simeq2 MeV for CONUS100. For CONUS+ the improvement stops at ∣gaγγ∣≃1.0×10−4|g_{a\gamma\gamma}| \simeq1.0 \times10^{-4} GeV−1^{-1} for ma≃0.35m_a \simeq0.35 MeV, and its contour closes at ma≃0.6m_a \simeq0.6 MeV, since heavier ALPs carry Ea≥maE_a \geq m_a above the DAQ range. At still larger masses and couplings, the reduction of the produced flux together with the decay of the ALPs before they reach the detector closes the sensitive region. The relative improvement of CONUS100 is not uniform in mass, since the signal composition changes between the scattering- and decay-dominated regimes.

The relevant experimental constraints in this mass range are set by the BaBar monophoton search [46] for ma≲60m_a \lesssim60 keV (∣gaγγ∣<1.7×10−4|g_{a\gamma\gamma}| < 1.7 \times10^{-4} GeV−1^{-1}) and, above, by beam-dump experiments, E137 [46, 47], E141 [48], NuCal [49], CHARM [50] and NA64 [51], as combined in Ref. [44]; the helioscope CAST [52, 53] dominates for ma≲200m_a \lesssim200 eV. Two further bounds are not drawn because they add no visible region in this window: NOMAD [54] (ma≲0.2m_a \lesssim0.2 keV) lies within 6% of BaBar, and the recast of MiniBooNE beam-dump data [55] lies within 5% of the beam-dump bound below 10 MeV. Between 3 keV and ∼7\sim7 MeV the reactor search with the NEON NaI(Tl) detector [18] adds to these constraints; NEON is the strongest bound for 33 keV ≲ma≲1\lesssim m_a \lesssim1 MeV, reaching ∣gaγγ∣≃1.3×10−4|g_{a\gamma\gamma}| \simeq1.3 \times10^{-4} GeV−1^{-1} at 10 keV and 9×10−69 \times10^{-6} GeV−1^{-1} at 1.2 MeV. With b=10b = 10 DRU the projected CONUS reach lies inside the experimentally excluded region over the whole mass range; the closest approach is at ma≃0.5m_a \simeq0.5 MeV, where the CONUS100 projection is a factor ≃1.8\simeq1.8 above the NEON limit, while CONUS+ remains at least a factor ≃3.5\simeq3.5 above it (at ma≃0.1m_a \simeq0.1 MeV).

Electron–ALP coupling. The corresponding single-coupling sensitivity to gaeeg_{aee} is shown in Figure 16. Below the e+e−e^+e^- threshold the inverse Compton-like and axio-electric yields scale as gaee4g_{aee}^{4} and the sensitivities are nearly mass independent, resulting in approximately horizontal contours: ∣gaee∣≃1.2×10−5|g_{aee}| \simeq1.2 \times10^{-5} (8.5×10−68.5 \times10^{-6}) for inverse Compton-like scattering and 1.7×10−51.7 \times10^{-5} (1.2×10−51.2 \times10^{-5}) for axio-electric absorption at CONUS+ (CONUS100). The CONUS+ contours end at ma≃0.6m_a \simeq0.6 MeV, the upper end of its DAQ window. The smaller window also lowers the CONUS+ background, which is why its scattering reach is slightly better than with a window up to 10 MeV.

A qualitatively different structure appears once ma>2me≃1.022m_a > 2m_e \simeq1.022 MeV, where the decay a→e+e−a \to e^+e^- of Eq. (43) becomes kinematically accessible. The finite lifetime then modifies both the flux reaching the detector and the probability for an ALP to decay inside the germanium, and produces for CONUS100 a closed decay region between ma=2mem_a = 2m_e and ≃2.4\simeq2.4 MeV, extending down to ∣gaee∣≃9×10−8|g_{aee}| \simeq9 \times10^{-8}; CONUS+ has no such region, since the e+e−e^+e^- pair deposits more than its 0.6 MeV DAQ range. At small gaeeg_{aee}, too few ALPs decay inside the detector; at large gaeeg_{aee}, a large fraction of the produced particles decay before reaching it.

Among experimental bounds, solar-ALP searches with XENONnT [56] and EDELWEISS-III [57] exclude ∣gaee∣≳2×10−12|g_{aee}| \gtrsim2 \times10^{-12} for ma≲1m_a \lesssim1 keV. Above ma≃60m_a \simeq60 keV the laboratory bounds are taken from the reanalysis of leptophilic ALPs in Ref. [58] (the E137 electron beam dump and the NA64 missing-energy and visible-decay searches) and from Ref. [18] (the NEON reactor search, NA64, and the beam dumps E137 [47], E141 [48] and Bechis et al. [59]). The E774 [60] and Orsay [61] electron beam dumps are taken from Ref. [62], with gaee=e ϵg_{aee} = e\,\epsilon in terms of the coupling ϵ\epsilon used there; the Orsay region is drawn down to the lower end of the figure of that reference, ∣gaee∣≃6×10−6|g_{aee}| \simeq6 \times10^{-6}, below which E137 already excludes. The NA64 missing-energy search excludes ∣gaee∣≳(1–2.5)×10−6|g_{aee}| \gtrsim(1\text{--}2.5) \times10^{-6} for 0.10.1 MeV ≲ma≲2me\lesssim m_a \lesssim2m_e, NEON reaches ∣gaee∣≃3×10−6|g_{aee}| \simeq3 \times10^{-6} between 60 and 900 keV, and above 2me2m_e the beam dumps exclude 2×10−8≲∣gaee∣≲10−42 \times10^{-8} \lesssim|g_{aee}| \lesssim10^{-4}. The CONUS scattering contours above 60 keV and the whole a→e+e−a \to e^{+}e^{-} region therefore lie inside the excluded region. For 1 keV ≲ma≲60\lesssim m_a \lesssim60 keV none of the displayed limits covers the CONUS contours. This window reflects the mass ranges shown in Refs. [18, 58]: the NA64 missing-energy bound does not depend on mam_a for ma≪mem_a \ll m_e and is expected to extend to lower masses, so the window should not be read as a genuine sensitivity gain.

Nuclear production with photon detection. Figure 17 shows the projected sensitivity to the product ∣gaγγgann(1)∣|g_{a\gamma\gamma}g_{ann}^{(1)}| for the isovector benchmark of Eq. (66) and the dominant 2.230 MeV p(n,γ)dp(n,\gamma)d line. Detection proceeds through inverse Primakoff conversion and a→γγa \to\gamma\gamma decay. The coupling product is used as the presentation variable, while the separate dependence of propagation and decay on gaγγg_{a\gamma\gamma} is retained in the calculation: for each mass, the curve gives the smallest product for which some pair (gaγγ,gann(1))(g_{a\gamma\gamma},g_{ann}^{(1)}) with $g_{ann}^{(1)} \le 1 reaches the 90% C.L. threshold.

Projected 90% C.L. sensitivity to $|g_{a\gamma\gamma}g_{ann}^{(1)}|$ for ALPs from the 2.230 MeV $p(n,\gamma)d$ line

Figure 17. Projected 90% C.L. sensitivity to ∣gaγγgann(1)∣|g_{a\gamma\gamma}g_{ann}^{(1)}| for ALPs from the 2.230 MeV p(n,γ)dp(n,\gamma)d line, for CONUS100 (dashed) with b=10b=10 DRU in a 1 keV bin, unit efficiency up to 10 MeV (assumed) and gann(1)≤1g_{ann}^{(1)}\leq1; the line is outside the 0.6 MeV DAQ window of CONUS+. Shaded regions are experimental limits: TEXONO [17] (green) and Borexino [63] from inverse Primakoff conversion (magenta) and from a→γγa\to\gamma\gamma decays of solar ALPs (magenta, hatched; gann(1)≤1g_{ann}^{(1)}\leq1).

Only CONUS100 is shown: the 2.230 MeV signal lies above the 0.6 MeV DAQ range of CONUS+. At low masses the sensitivity is governed by the long-lived scattering contribution, ∣gaγγgann(1)∣≃2.2×10−9 GeV−1|g_{a\gamma\gamma}g_{\mathrm{ann}}^{(1)}| \simeq2.2 \times10^{-9}\ \mathrm{GeV}^{-1} for CONUS100. At larger masses, a→γγa \to\gamma\gamma decays take over and the reach improves to 1.9×10−12 GeV−11.9 \times10^{-12}\ \mathrm{GeV}^{-1} at ma≃1.7 MeVm_a \simeq1.7\ \mathrm{MeV}. As mam_a approaches the 2.230 MeV transition energy, the production phase space closes as pa3p_a^3 and the minimal product diverges; the sharp upward turn near the endpoint is therefore determined by the source kinematics rather than by the detector response.

The same isovector combination is constrained by TEXONO and by searches for the 5.5 MeV solar ALPs from p(d,3He)Ap(d,{}^3\mathrm{He})A, whose emission is controlled by g3AN≡gann(1)g_{3AN} \equiv g_{\mathrm{ann}}^{(1)}. TEXONO [17] obtained ∣gaγγgann(1)∣<7.7×10−9 GeV−1|g_{a\gamma\gamma}g_{\mathrm{ann}}^{(1)}| < 7.7 \times10^{-9}\ \mathrm{GeV}^{-1} for ma<105 eVm_a < 10^5\ \mathrm{eV}. The complete Borexino data set [63] gives ∣gAγg3AN∣≤2.3×10−11 GeV−1|g_{A\gamma}g_{3AN}| \leq2.3 \times10^{-11}\ \mathrm{GeV}^{-1} from inverse Primakoff conversion (mA<1 MeVm_A < 1\ \mathrm{MeV}) and, from A→γγA \to\gamma\gamma decays inside the detector, ∣gAγg3AN∣mA2≤1.6×10−2 GeV−1eV2|g_{A\gamma}g_{3AN}|m_A^2 \leq1.6 \times10^{-2}\ \mathrm{GeV}^{-1}\mathrm{eV}^2 for ALPs that reach the Earth. We draw the latter for g3AN≤1g_{3AN} \leq1, including the (pA/pγ)3(p_A/p_\gamma)^3 production factor and the Sun–Earth survival probability. The smallest products accessible to CONUS correspond to long-lived ALPs and are therefore already excluded by Borexino. The product representation, however, hides the region in which the ALP decays between the Sun and the Earth but survives the 20 m reactor baseline; this region is discussed below with the coupling planes of Figure 19.

Nuclear production with electron detection. The corresponding sensitivity to ∣gaeegann(1)∣|g_{aee}g_{\mathrm{ann}}^{(1)}| is displayed in Figure 18, with detection through inverse Compton-like scattering, axio-electric absorption, and a→e+e−a \to e^+e^- decay. As in the photon-detection case, the coupling product is used only as the plotting variable. For the scattering channels the product fully determines the yield in the long-lived regime. For the decay channel we show the lower branch of the sensitive region, obtained in the limit of small gaeeg_{aee} at fixed product, where Psurv→1P_{\mathrm{surv}} \to1 and Pdec∝gaee2P_{\mathrm{dec}} \propto g_{aee}^2; with gann(1)≤1g_{\mathrm{ann}}^{(1)} \leq1 this requires gaee≳10−15g_{aee} \gtrsim10^{-15}, deep in the long-lived regime.

Projected sensitivity to axion-like particles

Figure 18. Projected 90% C.L. sensitivity to ∣gaeegann(1)∣|g_{aee}g_{ann}^{(1)}| for ALPs from the 2.230 MeV p(n,γ)dp(n,\gamma)d line, for CONUS100 (dashed) with b=10b = 10 DRU in a 1 keV bin and efficiency 1 up to 10 MeV (assumed); the line is outside the 0.6 MeV DAQ window of CONUS+. Shaded regions are experimental limits: BGO [64] (lavender), Borexino [63] (magenta), TEXONO [17] (thin light-green band just above the blue contour) and the 57Fe{}^{57}\mathrm{Fe} solar-ALP searches [57, 65, 66, 67] (brown).

As for the photon channels, only CONUS100 is shown. In the long-lived regime the scattering channels reach ∣gaeegann(1)∣≃1.2×10−10|g_{aee}g_{\mathrm{ann}}^{(1)}| \simeq1.2 \times10^{-10} through axio-electric absorption and 2.7×10−102.7 \times10^{-10} through inverse Compton-like scattering for CONUS100, comparable to TEXONO [17] (1.3×10−101.3 \times10^{-10}, ma<1 MeVm_a < 1\ \mathrm{MeV}) and well above the bounds from the 5.5 MeV solar ALPs: Borexino [63], ∣gAeg3AN∣≤1.9×10−13|g_{A e}g_{3AN}| \leq1.9 \times10^{-13} (mA<1 MeVm_A < 1\ \mathrm{MeV}), and BGO [64], ∣gAeg3AN∣<1.9×10−10|g_{A e}g_{3AN}| < 1.9 \times10^{-10} (quoted for mA→0m_A \to0; its kinematic weakening is of a few per cent up to 1 MeV). Below 14.4 keV, the 57^{57}Fe solar-ALP searches (EDELWEISS-III [57], Majorana Demonstrator [65], CDEX-1B [66], PandaX-II [67]) constrain ∣gaeegann(1)∣≲4×10−17|g_{aee}g_{\mathrm{ann}}^{(1)}| \lesssim4 \times10^{-17}, which coincides with ∣gaeegann∣|g_{aee}g_{\mathrm{ann}}| for gann(1)=0g_{\mathrm{ann}}^{(1)} = 0.

The high-mass structure is controlled by two physical thresholds. The a→e+e−a \to e^+e^- channel opens at ma=2mem_a = 2m_e, and as mam_a approaches the 2.230 MeV transition energy the production phase space vanishes and all three detection channels deteriorate sharply. Between the two thresholds the decay channel produces a U-shaped sensitivity with minimum 1.3×10−151.3 \times10^{-15} at ma≃1.68 MeVm_a \simeq1.68\ \mathrm{MeV} for CONUS100. The published Borexino and BGO bounds are quoted for mA<1 MeVm_A < 1\ \mathrm{MeV}, so this region is not covered by existing experimental limits.

Coupling planes at fixed mass. Because the propagation and decay probabilities depend on the detection coupling separately, the product plots of Figure 17 and Figure 18 show only the smallest product that can be probed. Following Ref. [10], Figure 19 shows the full sensitivity in the (gaγγ,gann(1))(g_{a\gamma\gamma},g_{ann}^{(1)}) and (gaee,gann(1))(g_{aee},g_{ann}^{(1)}) planes for ma=1.2m_a=1.2 MeV, where decays dominate, and ma=10m_a=10 keV, where scattering dominates. In addition to the product bounds discussed above, we show the SNO search for 5.5 MeV solar ALPs detected through deuteron dissociation [45], which constrains the isovector coupling alone, ∣gaN3∣<2×10−5 GeV−1|g_{aN}^{3}|<2\times10^{-5}\ \mathrm{GeV}^{-1} (95% C.L.), i.e. ∣gann(1)∣=mN∣gaN3∣≲1.9×10−5|g_{ann}^{(1)}|=m_N|g_{aN}^{3}|\lesssim1.9\times10^{-5}, using the pseudoscalar form of the interaction given in that reference, together with the single-coupling laboratory and reactor bounds on gaγγg_{a\gamma\gamma} and gaeeg_{aee} of Figure 15 and Figure 16 at the corresponding mass. Since all the solar bounds require the ALP to reach the Earth, they are weighted with the Sun–Earth survival probability evaluated with the detection coupling.

At ma=10m_a=10 keV the CONUS sensitivities are entirely contained in the experimentally excluded region in both planes. At ma=1.2m_a = 1.2 MeV, instead, the solar searches lose sensitivity once the ALP decays before reaching the Earth, whereas it still survives the reactor baseline. CONUS100 then probes a region not covered by current experimental limits for 10−7≲∣gaγγ∣/GeV−1≲9×10−610^{-7} \lesssim|g_{a\gamma\gamma}|/\mathrm{GeV}^{-1} \lesssim9 \times10^{-6} with ∣gann(1)∣≳3×10−7|g_{ann}^{(1)}| \gtrsim3 \times10^{-7}, bounded from above in gaγγg_{a\gamma\gamma} by the NEON and beam-dump limits, and, through a→e+e−a \to e^{+}e^{-} decays, for 3×10−11≲∣gaee∣≲2×10−83 \times10^{-11} \lesssim|g_{aee}| \lesssim2 \times10^{-8} with ∣gann(1)∣≳9×10−8|g_{ann}^{(1)}| \gtrsim9 \times10^{-8}, bounded from above in gaeeg_{aee} by the E137 beam dump. CONUS+, with its present 0.6 MeV DAQ range, does not reach these regions. These are the regions in which a reactor experiment can improve on existing experimental bounds; astrophysical constraints, not shown here, are expected to be relevant in the same region.

Taken together, the ALP projections illustrate the complementarity between continuous-

spectrum and monochromatic reactor production. The former enables single-coupling searches in gaγγg_{a\gamma\gamma} and gaeeg_{aee}, while the latter provides sensitivity to products involving the nucleon coupling, ∣gaγγgann(1)∣\left|g_{a\gamma\gamma}g_{ann}^{(1)}\right| and ∣gaeegann(1)∣\left|g_{aee}g_{ann}^{(1)}\right|. The different structures of the projected curves can be traced to a small number of physical effects: finite ALP lifetimes, the opening of visible decay channels, and the closure of the available production phase space near the corresponding kinematic endpoints. When compared with current experimental limits only, the single-coupling and minimal-product projections lie inside regions that are already excluded, while two regions remain open: the a→e+e−a \to e^{+}e^{-} window of ∣gaeegann(1)∣\left|g_{aee}g_{ann}^{(1)}\right| between 2me2m_e and 2.2302.230 MeV, and, in the coupling planes at MeV masses, the region of ALPs that decay between the Sun and the Earth but survive the reactor baseline.

Conclusions

In this work we have investigated the sensitivity of the CONUS+ experiment and a future CONUS100 configuration to light weakly coupled bosons produced in a nuclear reactor. Two representative scenarios were considered: a kinetically mixed dark photon and axionlike particles coupled to photons, electrons, and nucleons. The short reactor–detector baseline, intense reactor photon field, and low-background germanium targets make these experiments sensitive to production and detection mechanisms that are complementary to conventional neutrino measurements.

For dark photons, we evaluated Compton-like production and inverse Compton-like detection while retaining the full finite-mass kinematics. The calculation also incorporates the total photon interaction cross section in uranium rather than the Compton-only approximation at few-hundred-keV energies. The resulting sensitivity is approximately mass independent in the low-mass regime, exhibits a moderate finite-mass variation at intermediate masses, and rapidly deteriorates as mA′m_{A^\prime} approaches the upper boundary of the accepted deposited-energy window, 0.60.6 MeV for CONUS+, while CONUS100, with a readout assumed to extend to 1010 MeV, remains sensitive up to 2me2m_e. For 0.50.5 keV ≲mA′≲0.5\lesssim m_{A^\prime} \lesssim0.5 MeV, above the reach of solar dark photon searches, the projections reach ϵ≃1.65×10−5\epsilon\simeq1.65 \times10^{-5} (CONUS+) and 1.2×10−51.2 \times10^{-5} (CONUS100), below the TEXONO limit as re-analysed in Ref. [22] by factors of 1.6 and 2.2.

For ALPs, we considered both continuous reactor-photon production and monochromatic production from nuclear magnetic transitions. The single-coupling searches involving gaγγg_{a\gamma\gamma} and gaeeg_{aee} display nearly mass-independent sensitivities in their long-lived, low-mass regimes, while finite lifetimes generate additional structure at larger masses. In particular, the opening of a→e+e−a \to e^{+}e^{-} at ma=2mem_a = 2m_e produces a characteristic decay region in the electron-coupling sensitivity. Searches based on p(n,γ)dp(n,\gamma)d production provide complementary access to the products ∣gaγγgann(1)∣\left|g_{a\gamma\gamma}g_{ann}^{(1)}\right| and ∣gaeegann(1)∣\left|g_{aee}g_{ann}^{(1)}\right|, with a sharp loss of sensitivity as mam_a approaches the 2.2302.230 MeV nuclear-transition endpoint.

Compared with current experimental limits only (laboratory, reactor and terrestrial solar-ALP searches), the single-coupling projections lie inside regions that are already excluded; the closest approach is for gaγγg_{a\gamma\gamma} at ma≃0.5m_a \simeq0.5 MeV, where CONUS100 is a factor ≃1.8\simeq1.8 above the NEON limit. The smallest coupling products, which correspond to longlived ALPs, are excluded by Borexino. Two regions, however, are not covered by published experimental limits: the a→e+e−a \to e^{+}e^{-} window of ∣gaeegann(1)∣\lvert g_{aee}g_{ann}^{(1)} \rvert between 2me2m_{e} and 2.230 MeV, reaching 1.3×10−151.3 \times10^{-15} for CONUS100, and, at MeV masses, the region of the (gaγγ,gann(1))(g_{a\gamma\gamma},g_{ann}^{(1)}) and (gaee,gann(1))(g_{aee},g_{ann}^{(1)}) planes in which the ALP decays between the Sun and the Earth but survives the 20 m reactor baseline. Astrophysical constraints, which were not considered here, are expected to be relevant in the same regions. Both regions require a readout window extending to MeV energies, as assumed here for CONUS100: with its present 0.6 MeV DAQ range, CONUS+ is sensitive only to the single-coupling channels below 0.6 MeV, which are already excluded.

Across all channels, the CONUS100 benchmark benefits substantially from the increase in target mass and integrated exposure. Its largest gain is not simply a rescaling of the CONUS+ result: the improvement depends on the signal-production mechanism, the detector channel, the ALP lifetime, and, for mixed-coupling searches, the separate coupling dependence of propagation and detection. The resulting projections therefore illustrate the broad physics reach that a larger reactor-based germanium experiment could achieve for weakly coupled particles.

The present sensitivities should nevertheless be interpreted as idealized projections. They use a common flat background benchmark of 10 DRU, simplified counting statistics, and unit signal efficiency within the corresponding analysis regions. A future experimental analysis should incorporate the measured energy-dependent background, detector response and containment efficiencies, reactor-on and reactor-off data, and a likelihood treatment of systematic and nuisance parameters. Implementing these ingredients will be necessary to convert the present physics reach estimates into detector-level exclusion limits.

Overall, the results show that reactor experiments optimized for low-energy neutrino measurements can also probe light dark-sector particles. With its present DAQ, CONUS+ can already improve on the TEXONO dark photon limit by a factor ≃1.6\simeq1.6, while the larger exposure and extended readout envisaged for CONUS100 give access to unexplored ALP parameter space at MeV masses.

Acknowledgments

We are grateful to the CONUS Collaboration for useful discussions. FSQ is supported by the Humboldt Foundation (Friedrich Wilhelm Bessel Research Award), ANID - Millennium Science Initiative Program - ICN2019_044, IIF-FINEP grant 213/2024, FAPESP Grants 2021/14335-0 and 2021/00449-4, and CNPq grants 307130/2021-5, 403521/2024-6, 309465/2026-5, and 446711/2025-0. LA acknowledges support from the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) under grants 88887.827404/2023-00 and 88881.128190/2025-01. LA also thanks the Department of Physics at the University of Oslo for the hospitality, support and useful discussions during the completion of this work. We also thank IIP for the local cluster bulletcluster, which was instrumental in this work. We acknowledge the use of Claude (Anthropic) for assistance in checking the numerical codes and in the language editing of the manuscript; the authors take full responsibility for the content of this work.

References

  1. [1]Planck Collaboration, Planck 2018 results. VI. Cosmological parameters, Astron. Astrophys. 641 (2020) A6 [1807.06209].
  2. [2]G. Arcadi, M. Dutra, P. Ghosh, M. Lindner, Y. Mambrini, M. Pierre et al., The waning of the WIMP? A review of models, searches, and constraints, Eur. Phys. J. C 78 (2018) 203 [1703.07364].DOI
  3. [3]G. Arcadi, D. Cabo-Almeida, M. Dutra, P. Ghosh, M. Lindner, Y. Mambrini et al., The Waning of the WIMP: Endgame?, Eur. Phys. J. C 85 (2025) 152 [2403.15860].DOI
  4. [4]M. Cirelli, A. Strumia and J. Zupan, Dark Matter, 2406.01705.arxiv.org/abs/2406.01705
  5. [5]J. Alexander et al., Dark Sectors 2016 Workshop: Community Report, 1608.08632.arxiv.org/abs/1608.08632
  6. [6]B. Holdom, Two U(1)’s and epsilon charge shifts, Phys. Lett. B 166 (1986) 196.DOI
  7. [7]H.K. Park, Detecting Dark Photons with Reactor Neutrino Experiments, Phys. Rev. Lett. 119 (2017) 081801 [1705.02470].DOI
  8. [8]I.G. Irastorza and J. Redondo, New experimental approaches in the search for axion-like particles, Prog. Part. Nucl. Phys. 102 (2018) 89 [1801.08127].DOI
  9. [9]L. Di Luzio, M. Giannotti, E. Nardi and L. Visinelli, The landscape of QCD axion models, Phys. Rept. 870 (2020) 1 [2003.01100].DOI
  10. [10]D. Aristizabal Sierra, V. De Romeri, L.J. Flores and D.K. Papoulias, Axionlike particles searches in reactor experiments, JHEP 03 (2021) 294 [2010.15712].arxiv.org/abs/2010.15712
  11. [11]M. Roos, Sources of Gamma Radiation in a Reactor Core, J. Nucl. Energy, Part B: Reactor Technology 1 (1959) 98.DOI
  12. [12]H. Bechteler, H. Faissner, H. Seyfarth and R. Yogeshwar, The spectrum of gamma radiation emitted in the FRJ-1 (Merlin) reactor core and moderator region, Tech. Rep. Juel-Spez 255, Institut fuer Kernphysik, Forschungszentrum Juelich (1984).
  13. [13]J.B. Dent, B. Dutta, D. Kim, S. Liao, R. Mahapatra, K. Sinha et al., New Directions for Axion Searches via Scattering at Reactor Neutrino Experiments, Phys. Rev. Lett. 124 (2020) 211804 [1912.05733].DOI
  14. [14]S. Weinberg, A New Light Boson?, Phys. Rev. Lett. 40 (1978) 223.DOI
  15. [15]T.W. Donnelly, S.J. Freedman, R.S. Lytel, R.D. Peccei and M. Schwartz, Do Axions Exist?, Phys. Rev. D 18 (1978) 1607.DOI
  16. [16]J.F. Cavaignac, A. Hoummada, D.H. Koang, B. Ost, B. Vignon, R. Wilson et al., A search for axions at a power reactor, Phys. Lett. B 121 (1983) 193.
  17. [17]TEXONO Collaboration collaboration, Search for axions from the Kuo-Sheng nuclear power reactor with a high-purity germanium detector, Phys. Rev. D 75 (2007) 052004 [hep-ex/0609001].
  18. [18]NEON collaboration, New Constraints on Axionlike Particles with the NEON Detector at a Nuclear Reactor, Phys. Rev. Lett. 134 (2025) 201002 [2406.06117].DOI
  19. [19]CONUS Collaboration collaboration, CONUS+ Experiment, Eur. Phys. J. C 84 (2024) 1265 [2407.11912].arxiv.org/abs/2407.11912
  20. [20]M. Lindner et al., CONUS100: Precision neutrino physics with coherent elastic scattering of reactor neutrinos, 2025.
  21. [21]M. Krcmar, Z. Krecak, A. Ljubicic, M. Stipcevic and D.A. Bradley, Search for solar axions using Li-7, Phys. Rev. D 64 (2001) 115016 [hep-ex/0104035].DOI
  22. [22]M. Danilov, S. Demidov and D. Gorbunov, Constraints on Hidden Photons Produced in Nuclear Reactors, Phys. Rev. Lett. 122 (2019) 041801 [1804.10777].DOI
  23. [23]M.J. Berger, J.H. Hubbell, S.M. Seltzer, J. Chang, J.S. Coursey, R. Sukumar et al., “XCOM: Photon Cross Section Database, Version 1.5.” NIST Standard Reference Database 8, 2010. 10.18434/T48G6X.
  24. [24]M. Du, J. Liu, X.-P. Wang and T. Wu, Amplifying nonresonant production of dark sector particles in scattering dominance regime, Phys. Rev. D 109 (2024) 055041 [2309.00231].DOI
  25. [25]S.V. Demidov, D.S. Gorbunov and A.L. Polonski, Comment on “Amplifying nonresonant production of dark sector particles in scattering dominance regime”, Phys. Rev. D 113 (2026) 098701 [2512.17652].arxiv.org/abs/2512.17652
  26. [26]D. Aloni, C. Fanelli, Y. Soreq and M. Williams, Photoproduction of Axionlike Particles, Phys. Rev. Lett. 123 (2019) 071801 [1903.03586].DOI
  27. [27]Particle Data Group collaboration, Review of Particle Physics, Phys. Rev. D 98 (2018) 030001.
  28. [28]Y.-S. Tsai, Axion bremsstrahlung by an electron beam, Phys. Rev. D 34 (1986) 1326.pubmed.ncbi.nlm.nih.gov/9957291
  29. [29]Y.-S. Tsai, Pair production and bremsstrahlung of charged leptons, Rev. Mod. Phys. 46 (1974) 815.DOI
  30. [30]R.H. Helm, Inelastic and Elastic Scattering of 187-MeV Electrons from Selected Even-Even Nuclei, Phys. Rev. 104 (1956) 1466.DOI
  31. [31]D. Aristizabal Sierra, J. Liao and D. Marfatia, Impact of form factor uncertainties on interpretations of coherent elastic neutrino-nucleus scattering data, JHEP 06 (2019) 141 [1902.07398].DOI
  32. [32]S.J. Brodsky, E. Mottola, I.J. Muzinich and M. Soldate, Laser-Induced Axion Photoproduction, Phys. Rev. Lett. 56 (1986) 1763.
  33. [33]R. Chanda, J.F. Nieves and P.B. Pal, Astrophysical constraints on axion and Majoron couplings, Phys. Rev. D 37 (1988) 2714.pubmed.ncbi.nlm.nih.gov/9958550
  34. [34]F.T. Avignone et al., Search for Axions From the 1115-keV Transition of 65Cu, Phys. Rev. D 37 (1988) 618.pubmed.ncbi.nlm.nih.gov/9958722
  35. [35]S. Dimopoulos, G.D. Starkman and B.W. Lynn, Atomic enhancements in the detection of axions, Mod. Phys. Lett. A 1 (1986) 491.DOI
  36. [36]F.T. Avignone et al., Laboratory limits on solar axions from an ultralow-background germanium spectrometer, Phys. Rev. D 35 (1987) 2752.pubmed.ncbi.nlm.nih.gov/9957984
  37. [37]M. Pospelov, A. Ritz and M.B. Voloshin, Bosonic super-WIMPs as keV-scale dark matter, Phys. Rev. D 78 (2008) 115012 [0807.3279].DOI
  38. [38]A. Derevianko, V.A. Dzuba, V.V. Flambaum and M. Pospelov, Axio-electric effect, Phys. Rev. D 82 (2010) 065006 [1007.1833].DOI
  39. [39]A. Barroso and N.C. Mukhopadhyay, Nuclear axion decay, Phys. Rev. C 24 (1981) 2382.DOI
  40. [40]W.C. Haxton and K.Y. Lee, Red giant evolution, metallicity, and new bounds on hadronic axions, Phys. Rev. Lett. 66 (1991) 2557.pubmed.ncbi.nlm.nih.gov/10043553
  41. [41]N.A. Ackermann, Coherent Elastic Antineutrino-Nucleus Scattering with CONUS+: Precision Background Modeling and Signal Extraction, phd thesis, Heidelberg University, Heidelberg, Germany, 2025.DOI
  42. [42]J. Jaeckel and S. Roy, Spectroscopy as a test of Coulomb’s law: A probe of the hidden sector, Phys. Rev. D 82 (2010) 125020 [1008.3536].DOI
  43. [43]H. An, M. Pospelov, J. Pradler and A. Ritz, New limits on dark photons from solar emission and keV scale dark matter, Phys. Rev. D 102 (2020) 115022 [2006.13929].DOI
  44. [44]C. O’Hare, “cajohare/AxionLimits: AxionLimits.” https://cajohare.github.io/AxionLimits/, 2020. 10.5281/zenodo.3932430.DOI
  45. [45]A. Bhusal, N. Houston and T. Li, Searching for Solar Axions Using Data from the Sudbury Neutrino Observatory, Phys. Rev. Lett. 126 (2021) 091601 [2004.02733].DOI
  46. [46]M.J. Dolan, T. Ferber, C. Hearty, F. Kahlhoefer and K. Schmidt-Hoberg, Revised constraints and Belle II sensitivity for visible and invisible axion-like particles, JHEP 12 (2017) 094 [1709.00009].
  47. [47]J.D. Bjorken et al., Search for neutral metastable penetrating particles produced in the SLAC beam dump, Phys. Rev. D 38 (1988) 3375.
  48. [48]E.M. Riordan et al., A search for short lived axions in an electron beam dump experiment, Phys. Rev. Lett. 59 (1987) 755.
  49. [49]J. Blümlein et al., Limits on neutral light scalar and pseudoscalar particles in a proton beam dump experiment, Z. Phys. C 51 (1991) 341.
  50. [50]CHARM collaboration, Search for axion like particle production in 400 GeV proton-copper interactions, Phys. Lett. B 157 (1985) 458.
  51. [51]NA64 collaboration, Search for axionlike and scalar particles with the NA64 experiment, Phys. Rev. Lett. 125 (2020) 081801 [2005.02710].DOI
  52. [52]CAST collaboration, An Improved limit on the axion-photon coupling from the CAST experiment, JCAP 04 (2007) 010 [hep-ex/0702006].DOI
  53. [53]CAST collaboration, New CAST Limit on the Axion-Photon Interaction, Nature Phys. 13 (2017) 584 [1705.02290].
  54. [54]NOMAD collaboration, Search for eV (pseudo)scalar penetrating particles in the SPS neutrino beam, Phys. Lett. B 479 (2000) 371 [hep-ex/0003028].
  55. [55]F. Capozzi, B. Dutta, G. Gurung, W. Jang, I.M. Shoemaker, A. Thompson et al., New constraints on ALP electron and photon couplings from ArgonNeut and the MiniBooNE beam dump, Phys. Rev. D 108 (2023) 075019 [2307.03878].arxiv.org/abs/2307.03878
  56. [56]XENON collaboration, Search for New Physics in Electronic Recoil Data from XENONnT, Phys. Rev. Lett. 129 (2022) 161805 [2207.11330].DOI
  57. [57]EDELWEISS collaboration, Searches for electron interactions induced by new physics in the EDELWEISS-III Germanium bolometers, Phys. Rev. D 98 (2018) 082004 [1808.02340].
  58. [58]F. Eberhart, M. Fedele, F. Kahlhoefer, E. Ravensburg and R. Ziegler, Leptophilic ALPs in Laboratory Experiments, JHEP 12 (2025) 055 [2504.05873].
  59. [59]D.J. Bechis, T.W. Dombeck, R.W. Ellsworth, E.V. Sager, P.H. Steinberg, L.J. Teasdale et al., Search for axion production in low-energy electron bremsstrahlung, Phys. Rev. Lett. 42 (1979) 1511.
  60. [60]A. Bross, M. Crisler, S.H. Pordes, J. Volk, S. Errede and J. Wrbanek, A search for shortlived particles produced in an electron beam dump, Phys. Rev. Lett. 67 (1991) 2942.pubmed.ncbi.nlm.nih.gov/10044598
  61. [61]M. Davier and H. Nguyen Ngoc, An unambiguous search for a light Higgs boson, Phys. Lett. B 229 (1989) 150.
  62. [62]NA64 collaboration, Search for pseudoscalar bosons decaying into e⁺e⁻ pairs in the NA64 experiment at the CERN SPS, Phys. Rev. D 104 (2021) L111102 [2104.13342].
  63. [63]BOREXINO collaboration, Search for high energy 5.5 MeV solar axions with the complete Borexino dataset, Eur. Phys. J. C 85 (2025) 1182 [2504.19135].DOI
  64. [64]A.V. Derbin et al., Search for axioelectric effect of solar axions using BGO scintillating bolometer, Eur. Phys. J. C 74 (2014) 3035 [1405.3782].DOI
  65. [65]MAJORANA collaboration, New limits on bosonic dark matter, solar axions, Pauli exclusion principle violation, and electron decay from the Majorana Demonstrator, Phys. Rev. Lett. 118 (2017) 161801 [1612.00886].DOI
  66. [66]CDEX collaboration, Improved limits on solar axions and bosonic dark matter from the CDEX-1B experiment using the profile likelihood ratio method, Phys. Rev. D 101 (2020) 052003 [1911.03085].
  67. [67]PandaX collaboration, Limits on Axion Couplings from the First 80 Days of Data of the PandaX-II Experiment, Phys. Rev. Lett. 119 (2017) 181806 [1707.07921].DOI

Paper details

vlnc:W_7cjkw957
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 1

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Claude (Anthropic)

View statement
Looking for related papers…
Contents