Beam Dump eXperiment (BDX)

• BDX is a Jefferson Lab electron beam-dump experiment designed to produce and detect sub-GeV dark-sector states using high-intensity electron interactions with a beam dump.
• It employs a segmented CsI(Tl) calorimeter with active veto layers and extensive shielding to isolate rare dark matter–induced electromagnetic showers from beam-related backgrounds.
• BDX-MINI provided critical validation of the detector technology and background simulations, paving the way for extended facility enhancements and broader secondary beam research.

The Beam Dump eXperiment (BDX) is a Jefferson Lab electron beam-dump experiment centered on the production of sub-GeV dark-sector states in the Hall-A dump and their detection in a downstream, shielded detector. In its current formulation, BDX is an approved search for light dark matter produced in interactions of a high-intensity electron beam with the dump, and it is also the enabling project for a new underground vault that could be extended into a broader Beamdump Facility for secondary muon, neutrino, and hypothetical light dark matter beams (Achenbach et al., 6 Oct 2025, Battaglieri et al., 2016).

1. Conception, proposal history, and approval

The earliest BDX formulation was a Letter of Intent for a $1\,\mathrm{m}^3$ segmented plastic scintillator detector placed downstream of a high-intensity Jefferson Lab dump, receiving up to $10^{22}$ electrons on target in a one-year period. That version emphasized sensitivity to dark-matter–nucleon elastic scattering at the level of a thousand counts per year, while dark-matter–electron elastic scattering and inelastic dark matter channels were expected below 10 counts per year after requiring electromagnetic showers to exceed a few hundred MeV (Collaboration et al., 2014).

The full proposal redefined the experiment around the Hall-A beam dump and a segmented CsI(Tl) calorimeter. In that configuration, BDX would receive up to $10^{22}$ electrons on target in 285 days and would be sensitive to elastic dark-matter–electron and inelastic dark-matter scattering at the level of 10 counts per year, with the neutrino irreducible background setting the limiting floor (Battaglieri et al., 2016). This transition marked a shift from a low-threshold plastic-scintillator concept to a calorimetric electromagnetic-shower search optimized for forward electron recoils.

Subsequent updates were driven by concerns about beam-on backgrounds. The 2017 update addressed those concerns by implementing the detailed BDX experimental geometry in FLUKA, comparing FLUKA directly to GEANT4 in regions of validity, and developing a plan to measure the muon flux downstream of the Hall-A dump in its current configuration (Battaglieri et al., 2017). The 2018 update, published in 2019, reported a common FLUKA+GEANT4 framework for the Hall-A beam-background test and the proposed experiment, optimized the selection cuts, and concluded that with 285 days of a parasitic run at 65 $\mu$A, corresponding to $10^{22}$ electrons on target, BDX would lower exclusion limits by one to two orders of magnitude in the parameter space of dark-matter coupling versus mass (Battaglieri et al., 2019).

The project later entered the approved phase. The Jefferson Lab white paper states that the Beam Dump eXperiment is approved by the Physics Advisory Committee of Jefferson Lab, that the BDX-MINI experiment successfully demonstrated the concept using a dumped 2.1-GeV electron beam, and that the underground vault required for BDX could be extended into a Beamdump Facility with minimal additional installations (Achenbach et al., 6 Oct 2025). A common misconception is that BDX remained only a proposal; the later literature instead treats it as approved infrastructure with an explicit facility-development role.

2. Dark-sector benchmarks and detection logic

The primary benchmark is a light mediator coupled through kinetic mixing. In the formulation used in later BDX studies, the relevant Lagrangian is

$$\mathcal{L} = -\frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} - \frac{1}{4}F_{\mu\nu}F^{\mu\nu} + \frac{\epsilon}{2} F'_{\mu\nu}F^{\mu\nu} + \frac{m_{A'}^2}{2}A'_\mu A'^\mu + g_D A'_\mu J^\mu_\chi + e A_\mu J^\mu_{\rm EM},$$

with $g_D \equiv \sqrt{4\pi \alpha_D}$. In the direct annihilation regime with $m_\chi < m_{A'}$, the annihilation rate scales as $$\langle \sigma v \rangle \propto \epsilon^2\,\alpha_D \left(\frac{m_\chi}{m_{A'}}\right)^4$$, which motivates the standard BDX thermal-target variable

$$y \equiv \epsilon^2\,\alpha_D \left(\frac{m_\chi}{m_{A'}}\right)^4.$$

This is the parameterization in which the experiment’s reach is usually presented (Battaglieri et al., 2017).

The beam-dump production mechanism is electron bremsstrahlung on nuclei,

$10^{22}$0

followed by invisible decay $10^{22}$1 when kinematically open. The resulting light-dark-matter flux is strongly forward, so a compact downstream detector can efficiently sample it. Later BDX updates summarize the detection yield schematically as

$10^{22}$2

with the acceptance controlled by geometry, shielding, thresholds, and analysis cuts (Battaglieri et al., 2017).

The benchmark space is broader than elastic dark matter. The same framework includes inelastic dark matter with an off-diagonal interaction $10^{22}$3, so that $10^{22}$4 can de-excite inside the detector through $10^{22}$5. It also includes leptophilic mediators, such as gauged $10^{22}$6, which couple directly to leptonic currents rather than through kinetic mixing (Battaglieri et al., 2017). Earlier BDX formulations also included $10^{22}$7–nucleon elastic or quasi-elastic scattering with $10^{22}$8 MeV visible thresholds, but the later CsI(Tl) implementation is primarily organized around high-energy electromagnetic-shower signatures from $10^{22}$9–electron scattering and in-detector visible decays (Collaboration et al., 2014).

3. Hall-A layout, shielding, and baseline detector

CEBAF provides an up-to-12 GeV electron beam for fixed-target experiments, while the nominal BDX dark-matter studies were developed for an 11 GeV Hall-A beam-dump configuration (Achenbach et al., 6 Oct 2025, Battaglieri et al., 2017). In the later background and design studies, the detector front face is located at 20.8 m downstream of the dump entrance, behind a passive iron shielding wall of about 6 m thickness and within a new underground bunker approximately 20 m downstream of the dump, below grade, with dirt overburden and preceded by the existing concrete bunker and the iron wall (Battaglieri et al., 2017).

The detector concept is a segmented electromagnetic calorimeter based on CsI(Tl) crystals read out by SiPMs and surrounded by two active veto layers, an Inner Veto and an Outer Veto. The signal signature is a high-energy electromagnetic shower from $10^{22}$0–$10^{22}$1 scattering in the calorimeter, with a baseline trigger threshold $10^{22}$2 MeV, little activity in the vetoes, and timing consistent with beam (Battaglieri et al., 2017). This detector logic is specific: BDX is not simply a passive counting experiment behind shielding, but a calorimetric search for a forward electromagnetic topology.

The detailed implementation evolved, but later documents summarize the full detector as approximately 800 CsI(Tl) crystals totaling roughly $10^{22}$3 and grouped into modules (Battaglieri et al., 2019). The 2016 proposal described 8 modules of $10^{22}$4 crystals, with a detector cross section of about $10^{22}$5, active length about 295 cm, and total active volume about $10^{22}$6 (Battaglieri et al., 2016). The same proposal placed a lead layer between the inner and outer vetoes and treated the full instrument as a hermetic, shielded calorimeter intended to operate at the neutrino irreducible floor.

This geometry is central to BDX’s method. A plausible implication is that the experiment’s sensitivity is determined less by raw detector mass than by the combination of forward acceptance, underground siting, and the ability to veto cosmic and residual charged activity without self-vetoing genuine electromagnetic showers.

4. Backgrounds, simulations, and validation

Later BDX work treated beam-related background control as the decisive technical question. The 2017 update implemented the full Hall-A dump and shielding geometry in FLUKA-2011.2c.5, with tuned biasing weights to accelerate rare-event sampling while preserving accuracy, and compared the results directly with GEANT4. At the dump exit, the muon energy spectra and energy–azimuth correlations agreed well within statistical uncertainties; for neutrons, the two codes agreed in the high-energy tail. The central conclusion was that for energies $10^{22}$7 MeV only neutrinos propagate through the planned shielding to the BDX detector (Battaglieri et al., 2017).

In that same framework, FLUKA tracked $10^{22}$8, $10^{22}$9, $\mu$0, and $\mu$1 to the detector front face, simulated $\mu$2–N and $\mu$3–N interactions inside CsI(Tl), and propagated the final states through a detailed GEANT4 detector model. The update reported that $\mu$4–$\mu$5 scattering is negligible, with an analytic estimate of fewer than one $\mu$6–$\mu$7 interaction in CsI(Tl) for $\mu$8 electrons on target, and that after cuts exploiting topology and kinematics the extrapolated neutrino background for $\mu$9 electrons on target is about 10 events:

$10^{22}$0

This is the sense in which BDX is described as reaching the neutrino floor for electron fixed-target experiments (Battaglieri et al., 2017).

The 2018 update added a direct Hall-A validation campaign. It used wells downstream of the dump and a compact BDX-Hodo detector consisting of one CsI(Tl) crystal sandwiched between segmented plastic scintillators. In the optimized full-detector analysis, the best-performing cut set was $10^{22}$1 MeV, $10^{22}$2, and $10^{22}$3; with those cuts the expected cosmic background over 285 days was $10^{22}$4 counts, the expected neutrino-induced background was about 5 events over $10^{22}$5 electrons on target, and the optimized shielding reduced residual beam-related muons to order unity (Battaglieri et al., 2019). The same test program reported that a 4.3 GeV background run with $10^{22}$6 electrons on target showed no beam-related counts above cosmics in the relevant energy range (Battaglieri et al., 2019).

A persistent misconception is that the final BDX signal region is dominated by muons, photons, or neutrons leaking from the dump. The later design literature explicitly rejects that picture: with the planned iron shielding and underground siting, neutrinos are treated as the only irreducible beam-related background above threshold (Battaglieri et al., 2017, Battaglieri et al., 2019).

5. BDX-MINI as pilot experiment

BDX-MINI was built as a reduced-scale implementation of the BDX logic. It used a PbWO$10^{22}$7 electromagnetic calorimeter surrounded by a passive tungsten layer and a hermetic veto system with Inner Veto and Outer Veto detectors made of EJ200 plastic scintillator, all read out with SiPMs. The apparatus was installed in Well-1, 26 m downstream of the Hall-A aluminum–water beam dump, behind 5.4 m of concrete and 14.2 m of dirt, and operated with the 2.176 GeV continuous-wave CEBAF beam (Battaglieri et al., 2020).

The pilot run accumulated $10^{22}$8 electrons on target in six months. Beam-off periods accounted for about 50% of data-taking, enabling direct characterization of the cosmic-ray background. The analysis was blind and used a maximum-likelihood framework with beam-on and beam-off samples, nuisance parameters for detector position, veto thresholds, and energy scale, and an energy-binned likelihood that improved the expected limits by about a factor of two relative to a pure counting analysis (Battaglieri et al., 2022).

The beam-related neutrino background was simulated with FLUKA and GENIE and amounted to $10^{22}$9 anti-coincidence events per electron on target with $$\mathcal{L} = -\frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} - \frac{1}{4}F_{\mu\nu}F^{\mu\nu} + \frac{\epsilon}{2} F'_{\mu\nu}F^{\mu\nu} + \frac{m_{A'}^2}{2}A'_\mu A'^\mu + g_D A'_\mu J^\mu_\chi + e A_\mu J^\mu_{\rm EM},$$0 MeV, corresponding to fewer than one neutrino event over the full exposure. The observed samples were $$\mathcal{L} = -\frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} - \frac{1}{4}F_{\mu\nu}F^{\mu\nu} + \frac{\epsilon}{2} F'_{\mu\nu}F^{\mu\nu} + \frac{m_{A'}^2}{2}A'_\mu A'^\mu + g_D A'_\mu J^\mu_\chi + e A_\mu J^\mu_{\rm EM},$$1, $$\mathcal{L} = -\frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} - \frac{1}{4}F_{\mu\nu}F^{\mu\nu} + \frac{\epsilon}{2} F'_{\mu\nu}F^{\mu\nu} + \frac{m_{A'}^2}{2}A'_\mu A'^\mu + g_D A'_\mu J^\mu_\chi + e A_\mu J^\mu_{\rm EM},$$2, with $$\mathcal{L} = -\frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} - \frac{1}{4}F_{\mu\nu}F^{\mu\nu} + \frac{\epsilon}{2} F'_{\mu\nu}F^{\mu\nu} + \frac{m_{A'}^2}{2}A'_\mu A'^\mu + g_D A'_\mu J^\mu_\chi + e A_\mu J^\mu_{\rm EM},$$3 (Battaglieri et al., 2022). In the benchmark dark-photon scenario with $$\mathcal{L} = -\frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} - \frac{1}{4}F_{\mu\nu}F^{\mu\nu} + \frac{\epsilon}{2} F'_{\mu\nu}F^{\mu\nu} + \frac{m_{A'}^2}{2}A'_\mu A'^\mu + g_D A'_\mu J^\mu_\chi + e A_\mu J^\mu_{\rm EM},$$4 and $$\mathcal{L} = -\frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} - \frac{1}{4}F_{\mu\nu}F^{\mu\nu} + \frac{\epsilon}{2} F'_{\mu\nu}F^{\mu\nu} + \frac{m_{A'}^2}{2}A'_\mu A'^\mu + g_D A'_\mu J^\mu_\chi + e A_\mu J^\mu_{\rm EM},$$5, the best reach occurred near $$\mathcal{L} = -\frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} - \frac{1}{4}F_{\mu\nu}F^{\mu\nu} + \frac{\epsilon}{2} F'_{\mu\nu}F^{\mu\nu} + \frac{m_{A'}^2}{2}A'_\mu A'^\mu + g_D A'_\mu J^\mu_\chi + e A_\mu J^\mu_{\rm EM},$$6–4.5 MeV, where the fermionic case attained $$\mathcal{L} = -\frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} - \frac{1}{4}F_{\mu\nu}F^{\mu\nu} + \frac{\epsilon}{2} F'_{\mu\nu}F^{\mu\nu} + \frac{m_{A'}^2}{2}A'_\mu A'^\mu + g_D A'_\mu J^\mu_\chi + e A_\mu J^\mu_{\rm EM},$$7 at $$\mathcal{L} = -\frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} - \frac{1}{4}F_{\mu\nu}F^{\mu\nu} + \frac{\epsilon}{2} F'_{\mu\nu}F^{\mu\nu} + \frac{m_{A'}^2}{2}A'_\mu A'^\mu + g_D A'_\mu J^\mu_\chi + e A_\mu J^\mu_{\rm EM},$$8 MeV and the scalar case was slightly stronger at $$\mathcal{L} = -\frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} - \frac{1}{4}F_{\mu\nu}F^{\mu\nu} + \frac{\epsilon}{2} F'_{\mu\nu}F^{\mu\nu} + \frac{m_{A'}^2}{2}A'_\mu A'^\mu + g_D A'_\mu J^\mu_\chi + e A_\mu J^\mu_{\rm EM},$$9 MeV (Battaglieri et al., 2022).

BDX-MINI did not supersede the full experiment; it validated the detector technology, veto strategy, DAQ, and background treatment under realistic dump conditions. That role is important: in the BDX literature, BDX-MINI is consistently treated as a proof-of-principle rather than as an alternative physics program.

6. Facility extensions and related research directions

The 2025 Jefferson Lab white paper broadens BDX from a single dark-matter experiment into a beamdump-facility concept. It highlights intense secondary muon, neutrino, and hypothetical light dark matter beams produced in interactions of high-intensity electron beams with beam dumps, and states that the BDX underground vault could host additional instrumentation with minimal additional installations (Achenbach et al., 6 Oct 2025). Within that framework, the summary emphasizes a muon program with energies from 100 MeV to a few GeV and the unique feature of fully polarized muons at GeV energies, with applications in precise QED studies in electron scattering, nucleon form factors, muon–electron scattering for beyond-the-Standard-Model sensitivity, applied muon tomography and elemental analysis, and research and development for muon-collider components. It also highlights a neutrino program based on a characteristic decay-at-rest energy spectrum, precision CE$g_D \equiv \sqrt{4\pi \alpha_D}$0NS, Weinberg-angle measurements, beyond-the-Standard-Model searches, and synergy with long-baseline programs such as DUNE, potentially using advanced low-threshold detectors including liquid Ar or novel scintillating materials (Achenbach et al., 6 Oct 2025).

The BDX geometry has also been repurposed in explicit beyond-the-benchmark studies. One example is the leptophilic-scalar analysis in which secondary muons from the 11 GeV Hall-A dump radiate a light scalar through $g_D \equiv \sqrt{4\pi \alpha_D}$1. In that treatment, the simulated muon yield in the 0.5–11 GeV range is $g_D \equiv \sqrt{4\pi \alpha_D}$2 muons per electron on target, corresponding to order $g_D \equiv \sqrt{4\pi \alpha_D}$3 GeV-energy muons over the full BDX exposure, and the projected BDX sensitivity improves over E137 by orders of magnitude in the scalar-to-muon coupling versus scalar-mass plane (Marsicano et al., 2018).

Another example is the visible-decay axion-like-particle program. A 2025 study modeled BDX as a 10.6 GeV continuous-wave electron beam on the Hall-A aluminum dump, with $g_D \equiv \sqrt{4\pi \alpha_D}$4 electrons on target, detector distance $g_D \equiv \sqrt{4\pi \alpha_D}$5 m, shielding $g_D \equiv \sqrt{4\pi \alpha_D}$6 m, detector frontal acceptance area $g_D \equiv \sqrt{4\pi \alpha_D}$7, and an energy threshold $g_D \equiv \sqrt{4\pi \alpha_D}$8 MeV. Including the full electromagnetic cascade increased the BDX flux by at least $g_D \equiv \sqrt{4\pi \alpha_D}$9 across all masses in both coupling benchmarks and yielded a coupling-reach improvement of about $m_\chi < m_{A'}$0–$m_\chi < m_{A'}$1 in the long-lived-particle regime; the resulting sensitivity extended to previously unexplored regions up to $m_\chi < m_{A'}$2 GeV (Patrone et al., 17 Sep 2025).

The BDX name has also migrated into conceptually related but institutionally distinct proposals, including $m_\chi < m_{A'}$3BDX-DRIFT for directional CE$m_\chi < m_{A'}$4NS at LBNF and ILC-BDX for sub-GeV dark matter at the International Linear Collider (Sierra et al., 2021, Asai et al., 2023). This suggests that the Jefferson Lab experiment has become a methodological template: a beam-dump source, a forward, shielded detector, and a program organized around the interplay of rare new-physics signals with neutrino-limited backgrounds.