Forward Physics Facility (FPF) at CERN

• FPF is a proposed underground facility at CERN designed to capture far-forward, low-pT collision products for innovative studies in neutrino physics, QCD, and beyond.
• It integrates a suite of specialized detectors—FASER2, FASERν2, FLArE, and FORMOSA—each employing unique technologies from magnetic spectrometry to liquid-argon TPC and emulsion layers.
• FPF will enable high-precision measurements of TeV neutrino interactions and constrain hadronic models, dark-sector phenomena, and nuclear effects through unprecedented event statistics.

The Forward Physics Facility (FPF) is a proposed underground experimental area at CERN for the High-Luminosity LHC (HL-LHC), designed to instrument the far-forward region along the ATLAS beam collision axis and to host a suite of complementary experiments—currently centered on FLArE, FASER$\nu$2, FASER2, and FORMOSA. In the FPF conception, the HL-LHC is not only a high-$p_T$ collider but also a source of intense, highly collimated beams of neutrinos, muons, hadrons, and potentially new weakly coupled particles, enabling a combined program in neutrino physics, QCD, astroparticle physics, dark sectors, and long-lived particles (Anchordoqui et al., 24 Mar 2025, Adhikary et al., 2024).

1. Scientific premise and far-forward phase space

The FPF is motivated by a basic asymmetry in hadron-collider phenomenology: most LHC collisions are not rare, central, high-$p_T$ events, but low-$p_T$, forward events with vastly larger rates. In one formulation of the forward-physics case, high-$p_T$ processes have cross sections in the range $${\cal O}({\rm fb}) \lesssim \sigma_{\rm high\text{-}p_T} \lesssim {\cal O}({\rm pb})$$, yielding roughly $$10^3 \lesssim N^{\rm events}_{\rm high\text{-}p_T} \lesssim 10^6$$ after $3~{\rm ab}^{-1}$, whereas low-$p_T$ forward production has $$\sigma_{\rm low\text{-}p_T} \sim {\cal O}(100~{\rm mb})$$ and $p_T$0 at the HL-LHC (Anchordoqui, 2022). The far-forward direction is therefore not a peripheral corner of phase space but a dominant one that general-purpose detectors largely miss because particles emerge through the beamline apertures.

The geometric collimation is extreme. For $p_T$1, the characteristic angle is $p_T$2, so at a distance of hundreds of meters the signal footprint is tiny (Anchordoqui, 2022). This same kinematic regime contains the highest-energy human-made neutrinos, forward light- and heavy-hadron decay products, and many weakly coupled states whose production is peaked at small transverse momentum and whose decay lengths are enlarged by large boosts. Representative HL-LHC forward meson yields quoted within $p_T$3 mrad of the beam axis are $p_T$4, $p_T$5, $p_T$6 mesons, and $p_T$7 mesons (Anchordoqui et al., 2021).

The immediate experimental motivation was sharpened by the Run-3 forward program. FASER and FASER$p_T$8, together with SND@LHC, established that collider neutrinos can be directly observed and that the far-forward region can be instrumented in a low-background environment. The FPF is the proposed HL-LHC continuation of that program at larger scale, with dedicated infrastructure rather than opportunistic use of existing tunnels (Anchordoqui, 2022).

2. Site, geometry, and evolution of the proposal

The FPF has been described in several closely related geometrical baselines as the project matured. Early facility studies presented a preferred site on the line-of-sight of the ATLAS interaction point, about $p_T$9 m west of the interaction point, at a depth of $p_T$0 m, shielded by more than $p_T$1 m of rock, with a proposed cavern $p_T$2 m long and $p_T$3 m wide, targeting pseudorapidities above roughly $p_T$4 (Soldin, 2023). Later planning documents describe an updated baseline about $p_T$5 m west of ATLAS, still at $p_T$6 m depth and with more than $p_T$7 m of rock shielding, but now with a facility roughly $p_T$8 m long and $p_T$9 m internal width (Soldin, 2024). A more engineering-oriented summary describes a $p_T$0 m long, $p_T$1 m wide cavern, an $p_T$2 m deep shaft, and a closest approach of $p_T$3 m to the LHC tunnel (Anchordoqui et al., 24 Mar 2025).

These descriptions are not mutually exclusive so much as successive stages of design refinement. Across them, the persistent architectural features are the same: a new underground cavern on the ATLAS collision-axis line of sight, heavy shielding to suppress ordinary beam-induced activity, and enough space to host multiple detectors with distinct technologies and signatures. Geological and civil-engineering studies have been correspondingly prominent. A dedicated $p_T$4 cm diameter, $p_T$5 m deep core sample at the proposed shaft site found no show-stoppers, and vibration studies concluded that excavation can proceed during beam operation (Anchordoqui et al., 24 Mar 2025). Radiation studies indicate that access during LHC running should be possible with restrictions, a substantial operational advantage over more radiation-intense experimental areas (Anchordoqui et al., 24 Mar 2025).

A recurrent design distinction in the literature is between the current Run-3 forward installations and the FPF. FASER and FASER$p_T$6 operate in repurposed tunnel infrastructure. The FPF, by contrast, is a facility-scale proposal with shaft access, services, shielding, cryogenic accommodation, and enough hall volume to support simultaneous neutrino, LLP, and millicharge experiments (Anchordoqui et al., 2021).

3. Detector suite and experimental architecture

The current FPF baseline is organized around four complementary detectors, with earlier ecosystem studies also including AdvSND as an additional off-axis neutrino detector concept (Adhikary et al., 2024, Anchordoqui, 2022).

Experiment	Core technology	Stated role
FASER2	Magnetic spectrometer with decay volume	LLP decays; downstream charge and momentum analysis
FASER$p_T$72	Tungsten-emulsion detector	TeV neutrinos, especially $p_T$8, charm, and heavy-flavor topologies
FLArE	Liquid-argon TPC	Neutrino interactions and light dark matter scattering
FORMOSA	Scintillator/PMT array	Millicharged particles

FASER2 is the facility’s large spectrometer and LLP detector. In the current baseline it contains a $p_T$9 decay volume, tracking, calorimetry, iron absorber, and muon identification, with a baseline integrated magnetic field of $p_T$0 and tracker resolution of about $p_T$1. For this configuration, the quoted muon momentum resolution is approximately $p_T$2 at $p_T$3 and $p_T$4 at $p_T$5 (Anchordoqui et al., 24 Mar 2025). Besides LLP reconstruction, FASER2 is integral to the neutrino program because it can determine the charge of muons emerging from upstream neutrino interactions.

FASER$p_T$62 is the high-resolution on-axis neutrino target. The baseline design uses $p_T$7 emulsion layers interleaved with $p_T$8 mm tungsten plates, giving a $p_T$9 ton target with tungsten dimensions $${\cal O}({\rm fb}) \lesssim \sigma_{\rm high\text{-}p_T} \lesssim {\cal O}({\rm pb})$$0. The emulsion technology provides intrinsic position resolution around $${\cal O}({\rm fb}) \lesssim \sigma_{\rm high\text{-}p_T} \lesssim {\cal O}({\rm pb})$$1, making the detector suitable for $${\cal O}({\rm fb}) \lesssim \sigma_{\rm high\text{-}p_T} \lesssim {\cal O}({\rm pb})$$2 decays and charm/beauty vertexing (Anchordoqui et al., 24 Mar 2025). Because track densities above about $${\cal O}({\rm fb}) \lesssim \sigma_{\rm high\text{-}p_T} \lesssim {\cal O}({\rm pb})$$3 strain current analysis, annual film replacement is part of the operating concept (Anchordoqui et al., 24 Mar 2025).

FLArE is the liquid-argon fine-grained detector. In project-planning documents, the baseline cryostat is $${\cal O}({\rm fb}) \lesssim \sigma_{\rm high\text{-}p_T} \lesssim {\cal O}({\rm pb})$$4, with $${\cal O}({\rm fb}) \lesssim \sigma_{\rm high\text{-}p_T} \lesssim {\cal O}({\rm pb})$$5 TPC modules in a $${\cal O}({\rm fb}) \lesssim \sigma_{\rm high\text{-}p_T} \lesssim {\cal O}({\rm pb})$$6 arrangement and $${\cal O}({\rm fb}) \lesssim \sigma_{\rm high\text{-}p_T} \lesssim {\cal O}({\rm pb})$$7 separate $${\cal O}({\rm fb}) \lesssim \sigma_{\rm high\text{-}p_T} \lesssim {\cal O}({\rm pb})$$8 cm drift volumes. The quoted fiducial mass is about $${\cal O}({\rm fb}) \lesssim \sigma_{\rm high\text{-}p_T} \lesssim {\cal O}({\rm pb})$$9 tons, with about $$10^3 \lesssim N^{\rm events}_{\rm high\text{-}p_T} \lesssim 10^6$$0 tons active liquid argon, a nominal field of $$10^3 \lesssim N^{\rm events}_{\rm high\text{-}p_T} \lesssim 10^6$$1, and a preferred pixel pitch of $$10^3 \lesssim N^{\rm events}_{\rm high\text{-}p_T} \lesssim 10^6$$2 mm (Anchordoqui et al., 24 Mar 2025). In phenomenological studies, FLArE is often benchmarked more simply as a $$10^3 \lesssim N^{\rm events}_{\rm high\text{-}p_T} \lesssim 10^6$$3 ton detector with $$10^3 \lesssim N^{\rm events}_{\rm high\text{-}p_T} \lesssim 10^6$$4 transverse area and $$10^3 \lesssim N^{\rm events}_{\rm high\text{-}p_T} \lesssim 10^6$$5 exposure (Anchordoqui, 2022). Its stated niche is high-resolution calorimetric neutrino reconstruction together with visible recoil signatures from light dark matter.

FORMOSA is the dedicated millicharge detector. Current planning documents describe a $$10^3 \lesssim N^{\rm events}_{\rm high\text{-}p_T} \lesssim 10^6$$6 plastic-scintillator array read out by high-gain PMTs and operated through four-layer coincidence logic within a $$10^3 \lesssim N^{\rm events}_{\rm high\text{-}p_T} \lesssim 10^6$$7 ns time window, with timing resolution $$10^3 \lesssim N^{\rm events}_{\rm high\text{-}p_T} \lesssim 10^6$$8 ns (Adhikary et al., 2024). Earlier descriptions emphasized the same detection principle with somewhat smaller segmentation counts (Anchordoqui et al., 2021). Its role is narrow but distinctive: low-threshold sensitivity to tiny ionization signals from millicharged particles.

The detector suite is intentionally interdependent. FASER$$10^3 \lesssim N^{\rm events}_{\rm high\text{-}p_T} \lesssim 10^6$$92 and FLArE supply neutrino interaction targets with different reconstruction philosophies; FASER2 supplies charge and momentum information that neither of them can provide on its own; FORMOSA addresses a qualitatively different signature class. Earlier FPF ecosystem discussions also included AdvSND, especially for off-axis coverage $3~{\rm ab}^{-1}$0 and charm-sensitive neutrino measurements (Anchordoqui, 2022).

4. Neutrino, QCD, and astroparticle physics

The Standard Model core of the FPF is a high-statistics TeV neutrino program. Facility-level summaries describe expected samples of roughly $3~{\rm ab}^{-1}$1 electron-neutrino, $3~{\rm ab}^{-1}$2 muon-neutrino, and $3~{\rm ab}^{-1}$3 tau-neutrino interactions (Anchordoqui et al., 24 Mar 2025). An earlier detector-specific study of FASER$3~{\rm ab}^{-1}$42 for $3~{\rm ab}^{-1}$5 and $3~{\rm ab}^{-1}$6 TeV found strong generator dependence: $3~{\rm ab}^{-1}$7k/$3~{\rm ab}^{-1}$8k charged-current $3~{\rm ab}^{-1}$9 interactions, $p_T$0k/$p_T$1M charged-current $p_T$2, $p_T$3k/$p_T$4k charged-current $p_T$5, and $p_T$6k/$p_T$7k neutral-current interactions, where the paired values correspond to Sibyll- and DPMJET-based flux assumptions (Anchordoqui et al., 2021). These statistics underpin precision DIS at TeV energies, flavor-separated cross sections, and the first realistic collider-$p_T$8 program. Because FASER2 can measure muon charge downstream, the combined system is also expected to enable the first direct observation of anti-tau neutrinos in this environment (Anchordoqui et al., 24 Mar 2025).

The QCD program is inseparable from the neutrino program. Forward neutrino production is a proxy for forward hadron production, especially $p_T$9, $$\sigma_{\rm low\text{-}p_T} \sim {\cal O}(100~{\rm mb})$$0, hyperon, and charm sectors. In one summary of the DIS reach, the characteristic neutrino-scattering kinematics are written as $$\sigma_{\rm low\text{-}p_T} \sim {\cal O}(100~{\rm mb})$$1, and forward charm production relevant to the FPF probes gluon momentum fractions down to $$\sigma_{\rm low\text{-}p_T} \sim {\cal O}(100~{\rm mb})$$2 (Adhikary et al., 2024). This is central to small-$$\sigma_{\rm low\text{-}p_T} \sim {\cal O}(100~{\rm mb})$$3 PDF studies, forward charm phenomenology, and atmospheric prompt-neutrino modeling. The FPF is therefore not merely a consumer of hadronic models; it is designed to constrain them in a kinematic region that present collider data scarcely cover (Anchordoqui et al., 2021).

Nuclear effects are a separate line of inquiry. A GiBUU-based perspective study for $$\sigma_{\rm low\text{-}p_T} \sim {\cal O}(100~{\rm mb})$$4 $$\sigma_{\rm low\text{-}p_T} \sim {\cal O}(100~{\rm mb})$$5 on tungsten found that deep inelastic scattering dominates, with an invariant-mass distribution peaking around $$\sigma_{\rm low\text{-}p_T} \sim {\cal O}(100~{\rm mb})$$6. In the same calculation, final-state interactions on a tungsten target strongly reshaped baryonic observables: the baryon-multiplicity peak was reduced by about a factor of $$\sigma_{\rm low\text{-}p_T} \sim {\cal O}(100~{\rm mb})$$7, a long tail extended up to about $$\sigma_{\rm low\text{-}p_T} \sim {\cal O}(100~{\rm mb})$$8 baryons, and the proton kinetic-energy spectrum was depleted above $$\sigma_{\rm low\text{-}p_T} \sim {\cal O}(100~{\rm mb})$$9 while बढ़ strongly below $p_T$00 (Mosel et al., 2022). This motivates the use of the FPF as a laboratory for hadron formation, color transparency, and the neutrino EMC effect, not only for inclusive cross sections.

The astroparticle connection is equally central. Forward hadron production governs extensive air showers, atmospheric neutrino backgrounds, and the cosmic-ray muon puzzle. In FPF-focused astroparticle studies, predictions from different hadronic generators for forward $p_T$01 and $p_T$02 spectra through FASER$p_T$032 differ by up to a factor of two, substantially larger than expected statistical uncertainties (Soldin, 2024). A dedicated HL-LHC flux-comparison study went further: with $p_T$04 simulated $p_T$05 collisions per model and $p_T$06 normalization, it found typical model-to-model differences of $p_T$07–$p_T$08, larger still in the high-energy tail, and argued that current models overestimate the kaon-to-pion contribution ratio relative to recent FASER data (Soldin et al., 18 Sep 2025). In a toy strangeness-enhancement scenario, swapping pions into kaons at large pseudorapidity with probability $p_T$09 raises the peak $p_T$10 flux by a factor of $p_T$11 for $p_T$12 and by $p_T$13 for $p_T$14, with $p_T$15 partially accommodating Pierre Auger data in that model (Soldin, 2024). The same literature also ties the FPF to prompt atmospheric neutrinos, with the dominant contribution at $p_T$16 arising from charm produced at $p_T$17 (Soldin, 2023).

5. Beyond-the-Standard-Model program

The FPF BSM case is unusually broad because different forward detectors exploit different signatures: displaced decays, elastic or inelastic scattering, anomalous event-class ratios, low-threshold recoil electrons, and tiny ionization deposits. The general hidden-sector motivation is that light, weakly coupled particles are often both forward-produced and long-lived, making a downstream far-forward facility particularly effective (Anchordoqui, 2022).

For LLPs, the canonical targets are dark photons, dark Higgs bosons, heavy neutral leptons, and axion-like particles. In FPF-era studies, FASER2 is the principal decay spectrometer, with sensitivity arising from the large forward flux and long decay lever arm (Anchordoqui et al., 24 Mar 2025). FORMOSA extends this program to millicharged particles; one broad overview quotes sensitivity down to charges as low as $p_T$18 in the mass range $p_T$19 (Anchordoqui, 2022).

FLArE opens a different branch of dark-sector phenomenology: scattering signatures from stable or invisible-sector states. A dedicated study of electromagnetic form factors considered a Dirac fermion $p_T$20 with millicharge, magnetic dipole moment, electric dipole moment, anapole moment, or charge radius, produced through forward meson decays and Drell–Yan and detected by $p_T$21 in FLArE. In that analysis, FLArE provided new probes of sub-GeV dipole-coupled dark particles and strong millicharge sensitivity in the $p_T$22 to $p_T$23 range, while anapole and charge-radius interactions remained dominated by existing constraints (Kling et al., 2022).

Neutrino-sector mediators are another distinctive FPF target. A study of neutrino-portal dark matter with a neutrinophilic scalar $p_T$24 argued that future FPF detectors could probe mediator masses from about $p_T$25 to $p_T$26 GeV through charged-current scattering with large missing transverse momentum and through apparent excess $p_T$27 events. In that framework, the muon-plus-missing-$p_T$28 search was highly contingent on hadronic energy resolution at about the $p_T$29 level; degrading the hadronic resolution to $p_T$30 substantially weakened the reach (Kelly et al., 2021).

The facility has also been proposed as a precision probe of semileptonic and electroweak BSM effects in neutrino scattering. A leptoquark study of the vector singlet $p_T$31 scanned $p_T$32 and found that FLArE-100 gives the strongest reach among FASER$p_T$33, FASER$p_T$342, FLArE-10, and FLArE-100, with combined neutral- and charged-current sensitivities following the expected monotonic weakening with increasing mass (Cheung et al., 2023). A separate study of neutrino electromagnetic properties and $p_T$35 found that FPF detectors can improve laboratory constraints on the tau-neutrino magnetic moment and millicharge beyond DONUT by exploiting low-recoil electron scattering, and that FLArE could measure $p_T$36 to about $p_T$37 precision at $p_T$38 while also setting a world-leading bound on the electron-neutrino charge radius (Abraham et al., 2023).

More UV-specific constructions have also been mapped onto the FPF. In a Little String Theory realization of an anomalous bulk $p_T$39, FASER2 was argued to probe effective couplings $p_T$40 for $p_T$41, corresponding to string scales $p_T$42 under the model assumptions (Anchordoqui et al., 2022).

Neutrino trident scattering provides a final example of how the FPF neutrino program and BSM program intersect. In a study of the $p_T$43 model, the process $p_T$44 was treated as a benchmark BSM-sensitive channel for FASER$p_T$452 on tungsten. That analysis used a full $p_T$46 calculation without the equivalent photon approximation, assumed both coherent and incoherent nuclear contributions, and concluded that FASER$p_T$472 could extend coverage in the $p_T$48 plane beyond previous experiments and reach almost all of the region favored by the muon $p_T$49 anomaly (Francener et al., 2024).

6. Technical constraints, backgrounds, and project planning

The principal instrumental challenge for the neutrino detectors is the forward muon flux. A facility-wide background study based on FLUKA quoted about $p_T$50 within $p_T$51 cm of the line of sight at an instantaneous luminosity $p_T$52, a level judged acceptable but still motivating mitigation studies (Anchordoqui et al., 24 Mar 2025). A more detector-specific transport analysis for FASER$p_T$532, using SIBYLL generation, BDSIM beam transport, Geant4 tracking, and realistic field maps, framed the problem in terms of emulsion saturation: a tolerable track density of about $p_T$54 corresponds to a target muon density $p_T$55, or $p_T$56 for $p_T$57 per year. In that study the no-magnet flux at the detector was $p_T$58, an optimized magnet in the LHC tunnel alone reduced it to $p_T$59, and a three-stage system reached $p_T$60 (Ariga et al., 11 Jun 2026). The same study showed that geometry, multiple Coulomb scattering in rock, transverse magnet coverage, and installation constraints are as important as nominal field strength.

Project-planning documents present the FPF as technically mature but not yet fully approved. One planning summary describes the project as being at the pre-conceptual stage, with multiple internal technical notes on cost, integration, vibration, radiation, and scheduling, and with international strategy framing tied to the 2024–2026 European Particle Physics Strategy update (Adhikary et al., 2024). Cost accounting varies somewhat by aggregation convention. One recent breakdown quotes $p_T$61 MCHF for shaft and cavern civil construction, $p_T$62 MCHF for facility outfitting, and $p_T$63 MCHF for cryogenic infrastructure, for a total facility cost $p_T$64 MCHF, together with $p_T$65 MCHF in core experimental costs across the baseline suite (Adhikary et al., 2024). A closely related summary quotes about $p_T$66 MCHF for the facility and about $p_T$67 MCHF for the experimental program, excluding labor, contingency, and overhead (Anchordoqui et al., 24 Mar 2025). Civil engineering is expected to take about $p_T$68 years, and one stated planning line envisions funding ramp-up beginning in $p_T$69, detector commissioning by $p_T$70, and installation during Run 4 (Anchordoqui et al., 24 Mar 2025).

7. FCC extension and the scope of the term “FPF”

In present usage, “Forward Physics Facility” ordinarily denotes the HL-LHC facility along the ATLAS line of sight. A later extrapolation, explicitly called FPF@FCC, applies the same logic to a $p_T$71 TeV proton collider rather than redefining the HL-LHC FPF itself (Abraham et al., 2024). That distinction matters because some papers are strictly HL-LHC documents even when they mention the broader forward-physics idea. For example, the trident study centered on FASER$p_T$722 is explicitly an LHC-based FPF analysis; it does not present an FCC detector setup, an FCC flux model, or an FCC sensitivity projection (Francener et al., 2024).

FPF@FCC is nonetheless a natural extension of the concept. In a first quantitative FCC study, the assumed site is about $p_T$73 km downstream of the interaction point behind about $p_T$74 m of rock and concrete, with $p_T$75 at $p_T$76 TeV (Abraham et al., 2024). The baseline FCC$p_T$77 detector, sized $p_T$78, was projected to record $p_T$79M electron-flavor, $p_T$80M muon-flavor, and $p_T$81M tau-flavor charged-current interactions, while deeper or wider variants push the total program toward $p_T$82 detected electron/muon neutrinos and $p_T$83 tau neutrinos (Abraham et al., 2024). The same study extended the PDF and nuclear-dynamics reach to $p_T$84 in proton–lead running, projected charge-radius sensitivity down to the Standard Model scale for $p_T$85 and $p_T$86 and to about five times the SM value for $p_T$87, and found LLP sensitivity to masses as large as $p_T$88 GeV with couplings as small as $p_T$89, together with quirk masses up to $p_T$90 TeV (Abraham et al., 2024).

The relationship between the HL-LHC FPF and FPF@FCC is therefore sequential rather than interchangeable. The HL-LHC facility is the defined project under current planning; the FCC version is a future extension that uses the HL-LHC experience as both precedent and optimization guide (Abraham et al., 2024).