betaFeedback

WorksPhysics and Astronomy

First-Principles Nuclear Modeling for Light Dark Matter Experiments at the Intensity Frontier

Abstract

Accelerator-based experiments at the intensity frontier, in which a high-energy beam impinges on a nuclear target, serve as powerful probes of the light dark matter paradigm. Such experiments require precise modeling of the target nucleus for reliable signal predictions. We present the application of a many-body ab initio method to calculate light dark matter mediator production signal rates at electron fixed-target experiments, using chiral effective field theory interactions. Considering both elastic and quasi elastic scattering, we compute cross sections using a Monte Carlo event generator implementation informed by ab initio nuclear elastic form factors and spectral functions for three representative nuclei, 20^{20}Ne, 34^{34}Si, and 56^{56}Fe, at varying electron beam energies. We compare our results to a commonly used phenomenological parameterization, finding an increased signal yield by up to two orders of magnitude with our quasi elastic treatment and an agreement for lighter mediators.

Introduction. The challenge of unveiling the fundamental nature of dark matter (DM) is an interdisciplinary effort, drawing on tools and insight from across physics. Decades of null results from direct detection searches targeting GeV-scale Weakly Interacting Massive Particles (WIMPs) have sharpened the case for light DM, a mass range in which thermal freeze-out can also generate the observed relic abundance provided a light mediator connects the dark and visible sectors [1, 2, 3]. Traditional nuclear-recoil direct detection loses sensitivity for sub-GeV DM masses, where the energy transferred to a nucleus falls below typical detector thresholds [4], motivating accelerator-based approaches in which the DM or its mediator is produced directly.

Accelerator-based intensity frontier experiments provide a powerful probe of light new physics [1, 5]. Among these, electron beam fixed-target experiments, in which a high-intensity electron beam impinges on a stationary nuclear target, offer sensitivity to light dark sector particles produced via bremsstrahlung-like radiation off the beam electrons [6]. Several such experiments are currently running or proposed, including NA64 [7], LDMX [8, 9], DarkShine [10], Lohengrin [11], HPS experiment [12], and past searches including E137 [13], employing missing energy, missing momentum, and/or visible decay search strategies, spanning a range of beam energies.

Probing light DM at fixed-target experiments, however, requires precise control over nuclear physics, historically approximated rather than computed from more rigorous nuclear methods. Current analyses across this experimental program routinely treat the nuclear target through simplified parameterizations [6], neglecting the many-body nuclear structure known to shape the electron-nucleus response [14, 15, 16]. This approximation carries real consequences: robust discrimination of a signal excess from Standard Model backgrounds requires both the signal and background rates to be predicted with sufficient precision, and comparing results across experiments and setting exclusion limits requires predictions to be based on consistent and reliable nuclear physics. Moreover, as this class of experiments spans a range of target nuclei, an accurate treatment of nuclear structure offers a valuable tool for experimental design, allowing the choice of target material to be optimized for future searches.

Ab initio nuclear many-body methods provide systematically improvable predictions of ground- and excited-state properties of nuclear systems [17]. Rooted in quantum chromodynamics (QCD) through the chiral effective field theory (χ\chiEFT) interactions they employ [18], they have been proven capable of describing bulk properties of atomic nuclei over a wide region of the nuclear chart [19, 20]. Their reach, however, has long been restricted to (near-)spherical systems; only very recently has the concept of spontaneous symmetry breaking been exploited to extend them to deformed ones [21, 22, 23], which are characterized by strong long-range correlations and occupy the vast majority of the nuclear chart. While ab initio nuclear methods have been applied to DM direct detection [24, 25, 26], their use in modeling the target nuclei of intensity-frontier experiments remains unexplored.

In this Letter we present the first application of a deformed ab initio method, the deformed self-consistent Green’s function (dSCGF) approach [23], to the prediction of light dark matter (DM) signal rates at fixed-target experiments. Closely related variants of this approach have already been confronted with electron-nucleus scattering observables, yielding good agreement with experimental data for charge densities and form factors [27, 28] as well as for quasi elastic (QE) cross sections [15, 16]. We compute the dark mediator production cross section σ\sigma, which sets the expected signal yield N=σLN = \sigma L. Production proceeds through elastic scattering off the nucleus as a whole and QE scattering off individual nucleons. The elastic contribution is evaluated from the nuclear charge form factor obtained from the dSCGF charge distribution, while the QE contribution is treated in the impulse approximation [29], with the removal energy and momentum of the struck nucleon distributed according to the dSCGF spectral function [30].

Dark matter theory and production. While the formalism developed here applies to any light mediator connecting the visible and dark sectors, we focus on a vector mediator, the dark photon A′A^{\prime}, coupling both to DM and to Standard Model fields. The relevant Lagrangian describing the dark sector interactions is,

−LDM=Aμ′(ϵeJEMμ+gDJDμ),(1)-\mathcal{L}_{\mathrm{DM}} = A^{\prime}_{\mu}\left(\epsilon eJ_{\mathrm{EM}}^{\mu} + g_{D}J_{D}^{\mu}\right), \tag*{(1)}

where JEMμ≡∑fQffˉγμfJ_{\mathrm{EM}}^{\mu} \equiv\sum_{f} Q_{f}\bar{f}\gamma^{\mu}f is the electromagnetic current summed over charged Standard Model fermions ff with charge QfQ_{f}, JDμJ_{D}^{\mu} is the corresponding DM current, ee is the elementary charge, and ϵ\epsilon and gDg_{D} set the mediator’s coupling strength to ordinary matter and DM, respectively. Such a coupling to JEMμJ_{\mathrm{EM}}^{\mu} commonly arises from kinetic mixing between A′A^{\prime} and the photon [31, 32], though our analysis does not depend on the details of the underlying UV completion. We consider the sub-GeV mass range from 1 MeV to 1 GeV, where fixed-target experiments are most sensitive. We are agnostic to the mass ratio mA′/mDMm_{A^{\prime}}/m_{\mathrm{DM}}, the scale of gDg_{D}, and the specific structure of JDμJ_{D}^{\mu}, since we are only interested in the production rate of on-shell dark photon mediators, which could later decay visibly or invisibly.

Due to its coupling to electrons, dark photons can therefore be produced in nuclear fixed-target electron scattering through dark bremsstrahlung – the dark sector analog of ordinary bremsstrahlung. As shown in Figure 1, the process proceeds via exchange of a virtual photon carrying four-momentum q≡(ω,q)q \equiv(\omega,\mathbf{q}) between the electron and the nucleus, with the dark photon emitted from either the incoming or outgoing electron. The dark photon could also be emitted off the nucleus through virtual Compton scattering as studied in [33], however this diagram contributes sub-dominantly to the total cross section therefore we safely neglect it.

Figure 1. Feynman diagrams for dark photon production through dark bremsstrahlung, where a high-energy electron scatters off a nucleus (elastically or inelastically) and radiates a dark photon from the initial or final state.

The cross section of e−N→e−NA′e^{-}N \to e^{-}NA^{\prime} is governed by two independent kinematic variables: the three-momentum transfer ∣q∣|\mathbf{q}| and the energy transfer ω\omega of the virtual photon, with q2=−Q2=ω2−∣q∣2q^{2} = -Q^{2} = \omega^{2}-|\mathbf{q}|^{2} and q2<0q^{2}<0, where QQ is a scalar quantity. The dark photon mediator mass mA′m_{A^{\prime}} sets the minimum momentum transfer QQ accessible in the bremsstrahlung process, since producing an on-shell mediator of mass mA′m_{A^{\prime}} requires the target nucleus to absorb at least this much recoil. Light mediators are therefore radiated at small QQ, where the nucleus responds coherently, while heavy mediators require large QQ, resolving individual nucleons and pushing the process into the QE regime. Kinematically, ω<∣q∣\omega<|\mathbf{q}| (thus <Q<Q), so light mediators necessarily also probe small energy transfer ω\omega. At small ω\omega, thus small mediator masses, the cross section is dominated by elastic scattering, in which the nucleus remains in its ground state and recoils as a whole. The virtual photon couples coherently to the entire nuclear charge, so the elastic cross section is governed by the nuclear form factor – the Fourier transform of the ground state charge density. For larger ω\omega, thus larger mediator masses, the electron instead scatters inelastically with the nucleus. The dominant contribution is where the scattering occurs off the individual nucleons within the nucleus quasi-elastically [14]. The QE peak, broadened by Fermi motion and shifted by the nucleon separation energy, is centered at ω≈Q2/2mn\omega\approx Q^{2}/2m_{n}, where mnm_{n} is the nucleon mass. The cross section also receives contributions from inelastic transitions to discrete nuclear excited states and collective modes, which we neglect in this work.

In standard inclusive electron scattering, e−N→e−Ne^{-}N \to e^{-}N, the differential cross section factorizes into a leptonic tensor LμνL^{\mu\nu}, contracted with a nuclear tensor WμνW_{\mu\nu} that encodes all the nuclear structure information [34].

dσ∝LμνWμν,(2)d\sigma\propto L^{\mu\nu}W_{\mu\nu}, \tag*{(2)}

where the nuclear tensor can be decomposed into two structure functions, W1(∣q∣,ω)W_{1}(|\mathbf{q}|,\omega) and W2(∣q∣,ω)W_{2}(|\mathbf{q}|,\omega) [14] (see the End Matter for details on these structure functions). The dark bremsstrahlung process e−N→e−NA′e^{-}N \to e^{-}NA^{\prime} is governed by the same nuclear tensor, but with a different leptonic tensor [35]. We therefore implement the process exactly at the matrix-element level in MadGraph5_aMC@NLO [36], a Monte Carlo event generator for automated computation of cross sections. The nuclear structure functions are computed from the dSCGF ab initio many-body method and implemented via a custom Universal Feynman Output [37] model file for input to MadGraph, with details in the End Matter. We compute the dark bremsstrahlung cross section considering both elastic and QE scattering contributions, accounting for Pauli blocking in the latter. Electron screening is neglected; simplified atomic form factor estimates from Ref. [6] suggest a small effect at low mediator masses, though a treatment consistent with our ab initio many-body nuclear structure input is beyond the scope of this work.

Ab initio modeling of the target nucleus. The dSCGF ab initio method employed here is a self-consistent beyond-mean-field approach built on a deformed Hartree–Fock (dHF) reference state [38]. By breaking rotational symmetry, the dHF state efficiently captures the static long-range correlations that dominate in open-shell systems, while dSCGF adds dynamical correlations on top of it [39, 40]. The dSCGF expansion must be truncated at a fixed order, and this is typically done following the so-called Algebraic Diagrammatic Construction (ADC) scheme [41]. Calculations presented in this work are performed at second order, denoted as dSCGF(2), and break rotational-symmetry, such that the shape of the computed nuclei is axially-symmetric. A few third-order calculations, denoted as dSCGF(3), have been performed to estimate the many-body truncation uncertainty.

Two χ\chiEFT interactions are employed, ΔN2LOGO(394)\Delta\mathrm{N}^{2}\mathrm{LO}_{\mathrm{GO}}(394) [42] and N3LOTexas\mathrm{N}^{3}\mathrm{LO}_{\mathrm{Texas}} [43]. Matrix elements of the two- and three-nucleon (3N) parts of the Hamiltonian are generated with the NuHamil code [44] and expanded in a spherical harmonic-oscillator (sHO) basis, retaining all states up to emax⁡≡max⁡{2n+ℓ}=12e_{\max}\equiv\max\{2n+\ell\}=12, where nn and ℓ\ell denote the principal and orbital angular momentum quantum numbers. The sHO basis is characterized by an oscillator frequency ℏω\hbar\omega, chosen to optimize the convergence of the observables of interest in all calculations presented in this work. Three-body operators are included through the rank-reduction procedure of Ref. [45], and the corresponding matrix elements are truncated according to e3max⁡≤emax⁡(1)+emax⁡(2)+emax⁡(3)e_{3\max}\leq e_{\max}^{(1)}+e_{\max}^{(2)}+e_{\max}^{(3)}, the maximum total number of oscillator quanta carried by the three single-particle states. All calculations employ emax⁡=12e_{\max}=12 and e3max⁡=24e_{3\max}=24 (unless specified otherwise), for which the quantities reported here are converged at the subpercent level, making the associated uncertainty subdominant to the other sources discussed below.

A key advantage of working with a Green’s-function-based method is the straightforward access it grants to a broad set of ground- and excited-state observables through the Lehmann representation of the one-body propagator [46], which encodes the complete single-particle dynamics of the system:

gαβ(ω)=∑n⟨Ψ0A∣aα∣ΨnA+1⟩⟨ΨnA+1∣aβ†∣Ψ0A⟩ω−εn++iη+∑k⟨Ψ0A∣aβ†∣ΨkA−1⟩⟨ΨkA−1∣aα∣Ψ0A⟩ω−εk−−iη.(3)g_{\alpha\beta}(\omega)=\sum_n\frac{\langle\Psi_0^A|a_\alpha|\Psi_n^{A+1}\rangle\langle\Psi_n^{A+1}|a_\beta^\dagger|\Psi_0^A\rangle}{\omega-\varepsilon_n^++i\eta}+\sum_k\frac{\langle\Psi_0^A|a_\beta^\dagger|\Psi_k^{A-1}\rangle\langle\Psi_k^{A-1}|a_\alpha|\Psi_0^A\rangle}{\omega-\varepsilon_k^--i\eta}. \tag*{(3)}

where ∣Ψ0A⟩|\Psi_0^A\rangle is the ground state of the AA-nucleon system and ∣ΨnA+1⟩|\Psi_n^{A+1}\rangle, ∣ΨkA−1⟩|\Psi_k^{A-1}\rangle are eigenstates of the neighboring systems with one nucleon added or removed; Greek labels run over a complete orthonormal single-particle basis. The poles of the propagator are the one-nucleon separation energies, εn+≡EnA+1−E0A\varepsilon_n^+\equiv E_n^{A+1}-E_0^A and εk−≡E0A−EkA−1\varepsilon_k^-\equiv E_0^A-E_k^{A-1}, and the associated spectroscopic amplitudes, Xnα≡⟨ΨnA+1∣aα†∣Ψ0A⟩X_n^\alpha\equiv\langle\Psi_n^{A+1}|a_\alpha^\dagger|\Psi_0^A\rangle and Ykα≡⟨ΨkA−1∣aα∣Ψ0A⟩\mathcal{Y}_k^\alpha\equiv\langle\Psi_k^{A-1}|a_\alpha|\Psi_0^A\rangle, quantify the strength carried by each final state in the attachment and knockout channels.

Both quantities required in this work follow directly from these amplitudes. The hole spectral function entering the QE cross section reads

S(k,E)=∑f∣⟨nf∣⊗⟨ΨfA−1∣Ψ0A⟩∣2δ(E−E0A+EfA−1),(4)S(\mathbf{k},E)=\sum_f\left|\langle n_f|\otimes\langle\Psi_f^{A-1}|\Psi_0^A\rangle\right|^2\delta\left(E-E_0^A+E_f^{A-1}\right), \tag*{(4)}

which gives the joint probability of removing a nucleon of momentum k\mathbf{k} at removal energy EE from the ground state of the nucleus, leaving the residual system in the state ∣ΨfA−1⟩|\Psi_f^{A-1}\rangle, where the sum is over final states. Here ∣nf⟩|n_f\rangle denotes the single-particle state of the removed nucleon, carrying momentum k\mathbf{k}, which couples to the residual (A−1)(A-1)-body eigenstate to form the final state of the process. The one-body density is obtained as

ραβ=⟨Ψ0A∣aβ†aα∣Ψ0A⟩=∑k(Yβk)∗Yαk.(5)\rho_{\alpha\beta}=\langle\Psi_0^A|a_\beta^\dagger a_\alpha|\Psi_0^A\rangle=\sum_k(\mathcal{Y}_\beta^k)^*\mathcal{Y}_\alpha^k. \tag*{(5)}
Nucleusχ\chiEFT interactionkFpk_F^p [MeV]kFnk_F^n [MeV]
20Ne^{20}\mathrm{Ne}N3LOTexas\mathrm{N}^{3}\mathrm{LO}_{\mathrm{Texas}}173.46175.17
20Ne^{20}\mathrm{Ne}ΔN2LOGO(394)\Delta\mathrm{N}^{2}\mathrm{LO}_{\mathrm{GO}}(394)166.19168.37
34Si^{34}\mathrm{Si}N3LOTexas\mathrm{N}^{3}\mathrm{LO}_{\mathrm{Texas}}197.92149.70
34Si^{34}\mathrm{Si}ΔN2LOGO(394)\Delta\mathrm{N}^{2}\mathrm{LO}_{\mathrm{GO}}(394)197.62147.21
56Fe^{56}\mathrm{Fe}N3LOTexas\mathrm{N}^{3}\mathrm{LO}_{\mathrm{Texas}}175.90187.70
56Fe^{56}\mathrm{Fe}ΔN2LOGO(394)\Delta\mathrm{N}^{2}\mathrm{LO}_{\mathrm{GO}}(394)176.21188.12

Table 1. Proton and neutron Fermi momenta used in the Pauli-blocking suppression factor, Eq. (6), for each nucleus and χ\chiEFT interaction.

from which the charge density distribution is generated following Ref. [27]. Given the deformed setting of the many-body calculations, both the charge density and the spectral function in Eq. (4) are axially-symmetric.

In order to account for Pauli blocking, necessary for a correct description of the QE response at momentum transfers comparable to the Fermi momentum, we compute the suppression factor f(∣q∣)f(|\mathbf{q}|), whose analytical expression has been taken from Refs. [47, 48, 49] and reads

f(∣q∣)={34∣q∣kF−116(∣q∣kF)3,∣q∣<2kF,1,∣q∣≥2kF.(6)f(|\mathbf{q}|)=\begin{cases}\frac{3}{4}\frac{|\mathbf{q}|}{k_F}-\frac{1}{16}\left(\frac{|\mathbf{q}|}{k_F}\right)^3,&|\mathbf{q}|<2k_F,\\[6pt]1,&|\mathbf{q}|\geq2k_F.\end{cases} \tag*{(6)}

The Fermi momentum kFk_F was extracted as the momentum at which the momentum density distribution computed with DSCGF falls to half its value. The Fermi momentum for each nucleus and χ\chiEFT interaction considered in this work are presented in Table I for both protons and neutrons.

Results. In this Letter we consider three nuclei: 20Ne^{20}\mathrm{Ne}, 34Si^{34}\mathrm{Si}, and 56Fe^{56}\mathrm{Fe}. They span a broad range of masses, from light to medium-mass systems, as well as of intrinsic deformations: 34Si^{34}\mathrm{Si} is spherical, while 20Ne^{20}\mathrm{Ne} and 56Fe^{56}\mathrm{Fe} are strongly and weakly prolate, respectively. Figure 2 presents the elastic and QE contributions to dark photon mediator production via electron bremsstrahlung off 56Fe^{56}\mathrm{Fe}, computed using the Monte Carlo event generator MadGraph with the DSCGF method input. Below mA′∼0.1 GeVm_{A'}\sim0.1\,\mathrm{GeV}, with the exact value depending on the nucleus and beam energy, the elastic channel (in blue) dominates, above this mass, the QE channel (in magenta) dominates.

Figure 2. Dark photon mediator production cross section via bremsstrahlung with 56Fe^{56}\mathrm{Fe} as a function of the mediator mass mA′m_{A'}, for electron beam energies Ebeam=1E_{\mathrm{beam}} = 1, 1010, and 2020 GeV (left to right). Elastic (blue) and QE (magenta) contributions are evaluated with the dSCGF nuclear charge form factor and spectral function, respectively, using two χ\chiEFT interactions: N3^{3}LOTexas_{\mathrm{Texas}} (solid) and Δ\DeltaN2^{2}LOGO_{\mathrm{GO}}(394) (dashed). The analytic parametrization of Ref. [6], which combines elastic and inelastic nuclear form factors (Eqs. A18–A19 of Ref. [6]), is shown for reference (black). The inset plot in the middle panel compares the elastic and QE components of the Bjorken parametrization. σ\sigma is divided by ϵ2\epsilon^2, the overall kinetic mixing factor of Eq. (1).

Our results are compared against the widely used analytic parametrization of Ref. [6]. The elastic form factor is modeled in this parametrization as a smooth, monotonically decreasing function of QQ built from a phenomenological charge density, lacking any microscopic nuclear structure. Moreover, the QE channel is described by an elastic free proton form factor1, treating the QE scattering as coherent scattering off a single free proton scaled by ZZ, evaluated at fixed Q2Q^2 (lacking any ω\omega dependence). This treatment neglects Fermi motion broadening of the QE peak, nuclear binding, inter-nucleon correlations, the neutron contribution, and the effect of Pauli-blocking — all of which are simultaneously encoded in the spectral function formalism adopted here.

The ab initio elastic calculation agrees closely with the elastic parametrization presented in Ref. [6] for a certain momentum transfer regime. For QQ above ∼0.1\sim0.1 GeV, however, the form factor from the parametrization fails to provide the expected diffractive pattern [50, 51]. The discrepancies between these treatments of the form factor (as illustrated in the top row of Figure 5 in the End Matter) propagate directly into the cross section plotted here. As demonstrated in the inset of Figure 2, for mA′≳1m_{A'} \gtrsim1 GeV the two elastic treatments begin to diverge. The elastic channel from Ref. [6] parametrizes the electron screening, which suppresses the elastic form factor at small momentum transfer. Since it is absent in our formalism, at small mediator masses there is a small discrepancy with our elastic treatment.

Figure 5. Elastic form factors and QE structure functions for 20Ne^{20}\mathrm{Ne}, 34Si^{34}\mathrm{Si}, and 56Fe^{56}\mathrm{Fe}. Top row: elastic form factor squared, F(Q)2F(Q)^2, as a function of momentum transfer Q≡−q2Q \equiv\sqrt{-q^2}, comparing the phenomenological parametrization from Ref. [6] (black) with two χ\chiEFT interactions, N3LOTexas\mathrm{N}^{3}\mathrm{LO}_{\mathrm{Texas}} (blue, solid) and ΔN2LOGO(394)\Delta\mathrm{N}^{2}\mathrm{LO}_{\mathrm{GO}}(394) (dark blue, dashed). Bottom row: QE structure functions W1W_1 (green) and W2W_2 (pink) at fixed q=500q = 500 MeV, as functions of energy transfer ω\omega, computed from the N3LOTexas\mathrm{N}^{3}\mathrm{LO}_{\mathrm{Texas}} (solid) and ΔN2LOGO(394)\Delta\mathrm{N}^{2}\mathrm{LO}_{\mathrm{GO}}(394) (dashed) interactions via the dSCGF spectral function.

In the QE channel, by contrast, the agreement with Ref. [6] is far weaker. Using the ab initio treatment presented in this work, incorporating the nuclear structure functions plotted in the bottom row of Figure 5 in the End Matter in place of the analytic parametrization, the cross section lies systematically above the analytic cross section throughout the regime where the QE channel dominates the mediator production cross section. The enhancement is up to two orders of magnitude from the analytic curve. This discrepancy is not unexpected, since the analytic QE treatment is not informed by nuclear many-body correlations. The strikingly large size of the resulting enhancement signifies that a consistent many-body treatment of the full nuclear spectral function is necessary to obtain a reliable cross section in this regime. This translates directly into a higher projected signal yield and extends the accessible sensitivity to larger DM mediator masses beyond what the phenomenological parametrization of Ref. [6] would predict. For smaller mediator masses, by contrast, our treatment predicts a reduced QE cross section relative to the analytic parametrization. This reduction is attributed to Pauli blocking, which is absent from the analytic treatment, and our omission of the QE structure functions for ∣q∣<300|q| < 300 MeV: at such low momentum transfer the impulse approximation is not considered reliable and QE scattering is expected to be suppressed. We note that our results (magenta curve) are conservative with respect to the dark mediator signal yield, since including QE scattering beyond the ∣q∣|q| range considered here and other inelastic channels such as a Δ\Delta resonance production would further increase the inelastic cross section.

While χ\chiEFT interaction sensitivity has not been extensively studied, the two interactions used in this work, N3^{3}LOTexas_{\mathrm{Texas}} and Δ\DeltaN2^{2}LOGO_{\mathrm{GO}}(394) plotted as solid and dashed, respectively, in Figure 2, show compatible results. The same is found for the other two nuclei considered here. A few representative calculations have been performed at the dSCGF(3) level to estimate the impact of truncating the many-body expansion. This turns out to be the largest source of uncertainty among those considered here, amounting to a relative error of 3–5% on the total inelastic cross section and 1–10% on the elastic one, with the largest deviations occurring at the highest dark photon masses.

The cross sections σ/ϵ2\sigma/\epsilon^2 for the nuclei 20Ne^{20}\mathrm{Ne}, 34Si^{34}\mathrm{Si}, and 56Fe^{56}\mathrm{Fe} at Ebeam=10E_{\mathrm{beam}} = 10 GeV, are given in Figure 3, using the N3^{3}LOTexas_{\mathrm{Texas}} χ\chiEFT interaction. Increasing the nuclear mass enhances both channels, but with qualitatively different scaling. The elastic cross section is coherently enhanced over the ZZ protons, so heavier nuclei gain an approximately Z2Z^2 boost at fixed mA′m_{A'}. The QE channel, governed by the incoherent sum over single nucleon scattering weighted by the spectral function, scales only linearly with AA but retains support out to much larger mA′m_{A'}.

Figure 3. Dark photon mediator production cross section σ\sigma, with ϵ2\epsilon^2 factor divided out, via bremsstrahlung with the nuclei 20Ne^{20}\mathrm{Ne} (dark purple), 34Si^{34}\mathrm{Si} (violet), and 56Fe^{56}\mathrm{Fe} (teal) as a function of the mediator mass mA′m_{A'}. The electron beam energy is fixed to 10 GeV. Elastic (dashed) and QE (solid) contributions are evaluated with the dSCGF nuclear charge form factor and spectral function, respectively, using the N3LOTexas_{\mathrm{Texas}} χ\chiEFT interaction.

Summary and outlook. We demonstrate the imperative of interdisciplinary approaches bridging nuclear many-body theory and particle phenomenology to accurately make new physics predictions. In this work, we focus our analysis on light DM signatures at fixed-target facilities operating at the intensity frontier. We present cross sections for dark photon mediator production, incorporating state-of-the-art deformed many-body ab initio nuclear modeling. We embed our framework for both elastic and QE channels within MadGraph, integrating nuclear structure with Monte Carlo event generation.

The dSCGF many-body ab initio method is employed as our nuclear input, which enters the dark mediator production through elastic nuclear form factors and spectral functions, for a representative set of nuclei: 20Ne^{20}\mathrm{Ne}, 34Si^{34}\mathrm{Si}, and 56Fe^{56}\mathrm{Fe}. The sensitivity to the χ\chiEFT interaction is evaluated by performing calculations with two interactions, N3LOTexas_{\mathrm{Texas}} and ΔN2LOGO\Delta\mathrm{N}^{2}\mathrm{LO}_{\mathrm{GO}}(394), where we find agreement between both.

While the simple parametrization captures the elastic form factor in restricted kinematic windows, evaluating cross sections across diverse beam energies and target nuclei demands the complete, ab initio nuclear framework presented here. Our QE treatment, which includes the full spectral function formalism of the nuclear structure functions and the effect of Pauli-blocking, notably enhances the sensitivity to light DM for masses ≳0.1\gtrsim0.1 GeV relative to simplified phenomenological nuclear treatments. These results highlight the impact of such a framework for providing predictions for light DM mediator searches at intensity-frontier fixed-target experiments, establishing the bridge between systematically improvable first principles nuclear modeling and light DM phenomenology at accelerator-based searches.

Extending the predicted DM mediator event yield to new nuclear targets within our framework requires only a converged many-body calculation of that nucleus, with no additional phenomenological input. This ab initio treatment remains applicable well beyond the light- and medium-mass region considered here [23], making it possible to use these predictions directly to compute experimental sensitivities on DM model parameters and guide target material selection in proposed light DM fixed-target searches such as LDMX, Lohen­grin, DarkSHINE, and beyond.

Acknowledgments. The authors are grateful to Riccardo Catena and Xavier Roca-Maza for useful comments on the manuscript. T.G. has been funded by the Knut and Alice Wallenberg Foundation, and performed their research within the “Light Dark Matter” project (Dnr. KAW 2019.0080). A.S. acknowledges the Swedish Research Council (Grants No. 2021-04507 and No. 2025-05618). We acknowledge the National Academic Infrastructure for Supercomputing in Sweden (NAISS), funded by the Swedish Research Council, for providing computational resources.

End Matter

Nuclear structure functions for quasi elastic scattering. We make use of the impulse approximation, in which QE scattering reduces to a sum of scatterings off individual bound nucleons within the nucleus [14]. The nuclear spectral function in Eq. (4) directly enters the QE nuclear structure functions W1W_1 and W2W_2, through [14, 52, 53],

W1(∣q∣,ω)=∫d3k dE[ZSp(k,E)mnEk(w1p(∣q∣,ω~)+w2p(∣q∣,ω~)∣k×q∣22mn2∣q∣2)+(n)](7)W_1(|\mathbf{q}|,\omega)=\int d^3 k\,dE\left[Z S_p(\mathbf{k},E)\frac{m_n}{E_k}\left(w_1^p(|\mathbf{q}|,\widetilde{\omega})+w_2^p(|\mathbf{q}|,\widetilde{\omega})\frac{|\mathbf{k}\times\mathbf{q}|^2}{2m_n^2|\mathbf{q}|^2}\right)+(n)\right] \tag*{(7)}
W2(∣q∣,ω)=∫d3k dE[ZSp(k,E)mnEk(w1p(∣q∣,ω~)q2∣q∣2(q2q~2−1)+w2p(∣q∣,ω~)mn2(q4∣q∣4(Ek−ω~(Ekω~−k⋅q)q~2)2−q2∣k×q∣22∣q∣4)+(n)](8)W_2(|\mathbf{q}|,\omega)=\int d^3 k\,dE[Z S_p(\mathbf{k},E)\frac{m_n}{E_k}(w_1^p(|\mathbf{q}|,\widetilde{\omega})\frac{q^2}{|\mathbf{q}|^2}(\frac{q^2}{\widetilde{q}^2}-1)+\frac{w_2^p(|\mathbf{q}|,\widetilde{\omega})}{m_n^2}(\frac{q^4}{|\mathbf{q}|^4}(E_k-\frac{\widetilde{\omega}(E_k\widetilde{\omega}-\mathbf{k}\cdot\mathbf{q})}{\widetilde{q}^2})^2-\frac{q^2|\mathbf{k}\times\mathbf{q}|^2}{2|\mathbf{q}|^4})+(n)] \tag*{(8)}

where k\mathbf{k} and EkE_k are the initial momentum and energy of the struck nucleon, (n)(n) denotes the analogous neutron term weighted by A−ZA-Z instead of ZZ, and we take the w1Nw_1^N and w2Nw_2^N structure functions using the dipole and Galster parameterizations [54, 55]. The proton component of the spectral function, Sp(k,E)S_p(\mathbf{k},E), defined in Eq. (4), is represented in Figure 4 for dSCGF(2) and dSCGF(3). Increasing the order of the many-body truncation enhances the fragmentation of the peaks, in line with what has been observed in previous studies [39].

Proton spectral functions of $^{56}$Fe in the dSCGF(2) and dSCGF(3) approximations

Figure 4. Proton spectral function of 56^{56}Fe at kz=0 fm−1k_z=0\ {\rm fm}^{-1}, computed in the dSCGF(2) and dSCGF(3) approximations. Peaks are smeared with a Lorentzian of width Γ=1.5\Gamma=1.5 MeV for display purposes. Calculations employ the ΔN2LOGO(394)\Delta N^2\mathrm{LOGO}(394) interaction with emax⁡=6e_{\max}=6, e3max⁡=18e_{3\max}=18 and an oscillator frequency ℏω=18\hbar\omega=18 MeV.

We restrict the QE calculation to 300 MeV≲∣q∣≲800 MeV300\ {\rm MeV}\lesssim|\mathbf{q}|\lesssim800\ {\rm MeV}, the window in which the impulse approximation is expected to hold [53, 56]: large enough that the probe resolves individual nucleons, and small enough to remain below the onset of Δ\Delta and π\pi production channels [14]. The nucleon is bound, so only part of the energy transferred to the nucleus reaches the struck nucleon; this is accounted for by the shifted four-momentum q~≡(ω~,q)\widetilde{q}\equiv(\widetilde{\omega},\mathbf{q}), with ω~≡ω−E+mn−Ek\widetilde{\omega}\equiv\omega-E+m_n-E_k and the nucleon mass mnm_n [14].

The QE nuclear structure functions W1W_1 and W2W_2 for each nucleus and χ\chiEFT interaction are plotted as a function of energy transfer in the bottom row of Figure 5, for fixed ∣q∣=500|\mathbf{q}|=500 MeV. Notice that, similarly to the elastic form factor, both χ\chiEFT interactions coincide with each other. The nuclear structure functions grow with nucleon number, reflecting the increasing number of protons and neutrons available to absorb the momentum transfer.

Elastic nuclear form factor. The elastic nuclear form factor is computed from the charge density ρch(r)\rho_{\mathrm{ch}}(r) [27, 57],

F(Q2)=∫d3r ρch(r)eiq⋅r,(9)F(Q^2)=\int d^3r\,\rho_{\mathrm{ch}}(r)e^{i\mathbf{q}\cdot\mathbf{r}}, \tag*{(9)}

normalized such that F(0)=ZF(0)=Z, and since elastic scattering occurs at ω=0\omega=0 in the nuclear rest frame, the invariant four-momentum transfer q2q^2 reduces to −∣q∣2-|\mathbf{q}|^2, thus evaluating the Fourier transform with q\mathbf{q} yields the frame-independent F(Q2)F(Q^2). The squared elastic nuclear form factors are plotted in the top row of Figure 5 for each nucleus considered in this work. The ab initio form factors from the dSCGF many-body method (blue) deviate markedly from the standard parameterization of [6, 35] (black) for Q≳0.1Q\gtrsim0.1 GeV. This correction propagates directly through the cross section into the predicted DM signal yield. The two χ\chiEFT interactions considered in this work, N3N^3LOGO_{\mathrm{GO}} (blue, solid) and ΔN2\Delta N^2LOGO_{\mathrm{GO}}(394) (dark blue, dashed), yield similar form factors over the relevant momentum transfer range.

Event generator implementation. Both the elastic and QE channels are generated with MadGraph5_aMC@NLO using a custom UFO model, in which the nuclear physics enters through externally tabulated structure functions applied as event level reweighting.

For elastic scattering, within MadGraph, the nucleus is modeled as a spin-0 state of mass mNm_N, the nucleus mass, coupling to the photon through a point-like vertex. This vertex is then dressed with the nuclear form factor by reweighting each event with the form factor F(Q2)F(Q^2) from Eq. (9). The nucleus photon vertex is given by,

Γμel=ieF(Q2)(Pi+Pf)μ,(10)\Gamma^{\mathrm{el}}_{\mu}=ieF(Q^2)(P_i+P_f)_{\mu}, \tag*{(10)}

where PiP_i and PfP_f are the initial and final state nucleus momenta.

For QE scattering, the beam electron instead scatters with a bound nucleon within the nucleus and in the final state the nucleon is unbound. The target is represented by a free nucleon of mass mnm_n, the nucleon mass. The nuclear structure enters through the effective vertex,

ΓμQE=G1(∣q∣,ω)γμ+G2(∣q∣,ω)2mn(pf+pi)μ,(11)\Gamma^{\mathrm{QE}}_{\mu}=G_1(\lvert\mathbf{q}\rvert,\omega)\gamma_{\mu}+\frac{G_2(\lvert\mathbf{q}\rvert,\omega)}{2m_n}(p_f+p_i)_{\mu}, \tag*{(11)}

where pip_i and pfp_f are initial and final state nucleon momenta,

and with

G12=W1(∣q∣,ω)2mnωG22=2mnW1(∣q∣,ω)/ω−W2(∣q∣,ω)1+ω/2mn,(12)\begin{aligned} G_1^2&=\frac{W_1(\lvert\mathbf{q}\rvert,\omega)}{2m_n\omega}\\ G_2^2&=\frac{2m_nW_1(\lvert\mathbf{q}\rvert,\omega)/\omega-W_2(\lvert\mathbf{q}\rvert,\omega)}{1+\omega/2m_n}, \tag*{(12)} \end{aligned}

where the G12G_1^2 and G22G_2^2 contributions are generated as independent runs and summed at the cross section level, and W1W_1 and W2W_2 are computed from the spectral function described above. The form of ΓμQE\Gamma^{\mathrm{QE}}_{\mu} follows from the impulse approximation, where the matrix element is approximated by the free nucleon electromagnetic current operator evaluated between the bound initial nucleon and the ejected, on-shell final nucleon, with the initial nucleon’s momentum and removal energy distribution folded entirely into the spectral function. For an on-shell nucleon current, Lorentz invariance and current conservation (via the Gordon decomposition) restrict the vertex to two independent structures, γμ\gamma_{\mu} and (pf+pi)μ(p_f+p_i)_{\mu}. The coefficients G1G_1, G2G_2 are fixed by matching the squared amplitude built from ΓμQE\Gamma^{\mathrm{QE}}_{\mu} (averaging over final state spins and summing over initial state spins) to the general, Lorentz covariant, and parity and gauge invariant nuclear tensor WμνW_{\mu\nu}, Eq. (5) of [14] decomposition in terms of W1(∣q∣,ω)W_1(\lvert\mathbf{q}\rvert,\omega) and W2(∣q∣,ω)W_2(\lvert\mathbf{q}\rvert,\omega).

References

  1. [1]M. Battaglieri et al., in U.S. Cosmic Visions: New Ideas in Dark Matter (2017) arXiv:1707.04591 [hep-ph].
  2. [2]E. Izaguirre, G. Krnjaic, P. Schuster, and N. Toro, Phys. Rev. Lett. 115, 251301 (2015), arXiv:1505.00011 [hep-ph].DOI
  3. [3]S. Balan et al., JCAP 01, 053, arXiv:2405.17548 [hep-ph].DOI
  4. [4]R. Essig, J. Mardon, and T. Volansky, Phys. Rev. D 85, 076007 (2012), arXiv:1108.5383 [hep-ph].DOI
  5. [5]G. Krnjaic et al., A Snowmass Whitepaper: Dark Matter Production at Intensity-Frontier Experiments (2022), arXiv:2207.00597 [hep-ph].arxiv.org/abs/2207.00597
  6. [6]J. D. Bjorken, R. Essig, P. Schuster, and N. Toro, Phys. Rev. D 80, 075018 (2009), arXiv:0906.0580 [hep-ph].DOI
  7. [7]Y. M. Andreyev et al. (NA64), Searching for Light Dark Matter and Dark Sectors with the NA64 experiment at the CERN SPS (2025), arXiv:2505.14291 [hep-ex].
  8. [8]T. Åkesson et al. (LDMX) - The Light Dark Matter eXperiment (2025), arXiv:2508.11833 [hep-ex].arxiv.org/abs/2508.11833
  9. [9]A. Berger, R. Catena, J. Conrad, and T. R. Gray, Light Dark Matter Discovery Potential and Model Selection at LDMX (2026), arXiv:2607.24524 [hep-ph].arxiv.org/abs/2607.24524
  10. [10]J. Chen et al. (DarkSHINE), DarkSHINE Baseline Design Report: Physics Prospects and Detector Technologies (2024), arXiv:2411.09345 [physics.ins-det].
  11. [11]P. Bechtle et al., Eur. Phys. J. C 85, 600 (2025), arXiv:2410.10956 [hep-ex].arxiv.org/abs/2410.10956
  12. [12]N. Baltzell et al., The Heavy Photon Search Experiment (2022), arXiv:2203.08324 [hep-ex].arxiv.org/abs/2203.08324
  13. [13]B. Batell, R. Essig, and Z. Surujon, Phys. Rev. Lett. 113, 171802 (2014), arXiv:1406.2698 [hep-ph].DOI
  14. [14]O. Benhar, D. day, and I. Sick, Rev. Mod. Phys. 80, 189 (2008), arXiv:nucl-ex/0603029.DOI
  15. [15]N. Rocco and C. Barbieri, Phys. Rev. C 98, 025501 (2018).
  16. [16]C. Barbieri, N. Rocco, and V. Somà, Phys. Rev. C 100, 062501 (2019).
  17. [17]A. Ekström, C. Forssén, G. Hagen, G. R. Jansen, W. Jiang, and T. Papenbrock, Frontiers in Physics 11, 10.3389/fphy.2023.1129094 (2023).
  18. [18]R. Machleidt and F. Sammarruca, Phys. Scripta 91, 083007 (2016), arXiv:1608.05978 [nucl-th].DOI
  19. [19]S. R. Stroberg, J. D. Holt, A. Schwenk, and J. Simonis, Phys. Rev. Lett. 126, 022501 (2021).
  20. [20]A. Tichai, P. Demol, and T. Duguet, Phys. Lett. B 851, 138571 (2024).
  21. [21]G. Hagen, S. J. Novario, Z. H. Sun, T. Papenbrock, G. R. Jansen, J. G. Lietz, T. Duguet, and A. Tichai, Phys. Rev. C 105, 064311 (2022).
  22. [22]M. Frosini, T. Duguet, J.-P. Ebran, and V. Somà, Eur. Phys. J. A 58, 62 (2022), arXiv:2110.15737 [nucl-th].arxiv.org/abs/2110.15737
  23. [23]A. Scalesi, T. Duguet, and V. Somà, Deformed self-consistent Green’s function method for atomic nuclei at second and third order in the algebraic diagrammatic construction (2026), arXiv:2608.23700 [nucl-th].arxiv.org/abs/2608.23700
  24. [24]D. Gazda, R. Catena, and C. Forssén, Phys. Rev. D 95, 103011 (2017), arXiv:1612.09165 [hep-ph].DOI
  25. [25]B. S. Hu, J. Padua-Argüelles, S. Leutheuser, T. Miyagi, S. R. Stroberg, and J. D. Holt, Phys. Rev. Lett. 128, 072502 (2022), arXiv:2109.00193 [nucl-th].DOI
  26. [26]M. Hoferichter, P. Klos, J. Menéndez, and A. Schwenk, PoS CD2018, 095 (2019), arXiv:1903.11075 [hep-ph].arxiv.org/abs/1903.11075
  27. [27]T. Duguet, V. Somà, S. Lecluse, C. Barbieri, and P. Navrátil, Phys. Rev. C 95, 034319 (2017).
  28. [28]P. Arthuis, C. Barbieri, M. Vorabbi, and P. Finelli, Ab initio computation of charge densities for Sn and Xe isotopes (2020), arXiv:2002.02214 [nucl-th].DOI
  29. [29]O. Benhar, N. Farina, H. Nakamura, M. Sakuda, and R. Seki, Phys. Rev. D 72, 053005 (2005), arXiv:hep-ph/0506116.DOI
  30. [30]N. Rocco, A. Lovato, and O. Benhar, Phys. Rev. Lett. 116, 192501 (2016), arXiv:1512.07426 [nucl-th].DOI
  31. [31]B. Holdom, Phys. Lett. B 166, 196 (1986).
  32. [32]M. Fabbrichesi, E. Gabrielli, and G. Lanfranchi, The Dark Photon (2020), arXiv:2005.01515 [hep-ph].
  33. [33]M. Schürmann, H. K. Dreiner, and R. Gauld, Theory Calculations for LDMX and LOHENGRIN beyond Coherent Bethe-Heitler Scattering (2026), arXiv:2606.20327 [hep-ph].arxiv.org/abs/2606.20327
  34. [34]C. Itzykson and J. B. Zuber, Quantum Field Theory, International Series in Pure and Applied Physics (McGraw-Hill, New York, 1980).
  35. [35]K. J. Kim and Y.-S. Tsai, Phys. Rev. D 8, 3109 (1973).
  36. [36]J. Alwall, M. Herquet, F. Maltoni, O. Mattelaer, and T. Stelzer, JHEP 06, 128, arXiv:1106.0522 [hep-ph].DOI
  37. [37]L. Darmé et al., Eur. Phys. J. C 83, 631 (2023), arXiv:2304.09883 [hep-ph].DOI
  38. [38]P. Ring and P. Schuck, The Nuclear Many-Body Problem (Springer-Verlag, New-York, 1980).DOI
  39. [39]V. Somà, Front. in Phys. 8, 340 (2020), arXiv:2003.11321 [nucl-th].DOI
  40. [40]V. Somà, C. Barbieri, T. Duguet, and P. Navrátil, Eur. Phys. J. A 57, 135 (2021), arXiv:2009.01829 [nucl-th].DOI
  41. [41]F. Raimondi and C. Barbieri, Phys. Rev. C 97, 054308 (2018).
  42. [42]W. G. Jiang, A. Ekström, C. Forssén, G. Hagen, G. R. Jansen, and T. Papenbrock, Phys. Rev. C 102, 054301 (2020).
  43. [43]B. S. Hu, A. Ekström, C. Forssén, G. Hagen, W. G. Jiang, T. Miyagi, and T. Papenbrock, The neutron dripline in calcium isotopes from a chiral interaction (2025), arXiv:2512.11723 [nucl-th].DOI
  44. [44]T. Miyagi, Eur. Phys. J. A 59, 10.1140/epja/s10050-023-01039-y (2023).
  45. [45]M. Frosini, T. Duguet, B. Bally, J. Beaujeault-Taudière, J. P. Ebran, and V. Somà, Eur. Phys. J. A 57, 151 (2021), arXiv:2102.10120 [nucl-th].DOI
  46. [46]C. Barbieri and A. Carbone, in An Advanced Course in Computational Nuclear Physics, Lecture Notes in Physics, Vol. 936, edited by M. Hjorth-Jensen, M. Lombardo, and U. van Kolck (Springer, Cham, 2017).
  47. [47]J. S. Bell and M. J. G. Veltman, Phys. Lett. 5, 94 (1963).
  48. [48]P. Ballett, M. Hostert, S. Pascoli, Y. F. Perez-Gonzalez, Z. Tabrizi, and R. Zukanovich Funchal, JHEP 01, 119, arXiv:1807.10973 [hep-ph].DOI
  49. [49]A. Bodek, Pauli Blocking for a Relativistic Fermi Gas in Quasielastic Lepton Nucleus Scattering (2021), arXiv:2111.03633 [nucl-th].
  50. [50]B. Frois, J. B. Bellicard, J. M. Cavedon, M. Huet, P. Leconte, P. Ludeau, A. Nakada, P. Z. Hô, and I. Sick, Phys. Rev. Lett. 38, 152 (1977).
  51. [51]H. De Vries, C. W. De Jager, and C. De Vries, Atom. Data Nucl. Data Tabl. 36, 495 (1987).
  52. [52]T. De Forest, Nucl. Phys. A 414, 347 (1984).
  53. [53]N. Rocco, Front. in Phys. 8, 116 (2020).
  54. [54]S. Galster, H. Klein, J. Moritz, K. H. Schmidt, D. Wegener, and J. Bleckwenn, Nucl. Phys. B 32, 221 (1971).
  55. [55]C. F. Perdrisat, V. Punjabi, and M. Vanderhaeghen, Prog. Part. Nucl. Phys. 59, 694 (2007), arXiv:hep-ph/0612014.
  56. [56]J. Carlson, J. Jourdan, R. Schiavilla, and I. Sick, Phys. Rev. C 65, 024002 (2002), arXiv:nucl-th/0106047.DOI
  57. [57]T. De Forest, Jr. and J. D. Walecka, Adv. Phys. 15, 1 (1966).

Paper details

vlnc:W_sd4k54vk
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

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

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents