A proposed Fermilab rare-kaon program could generate about 10^18 to 10^20 kaons in an operating year, according to a new modeling study. The authors project that such samples could support precision tests of the CKM framework and searches for new physics. The figures are projections for a proposed facility, not measurements from a completed experiment.
The proposal centers on two decays: K+→π+νν̄ and K_L→π^0νν̄. The paper treats them as golden modes and uses them to project decay-rate precision, CKM unitarity-triangle constraints and sensitivity to high-energy new physics.
Four beam scenarios, very different usable samples
The researchers modeled proton beams at 3, 6.75, 8 and 120 GeV. They used FLUKA particle-transport simulations to follow what happens when the beam strikes a 10-centimetre gold cylindrical target, recording the K+ and K_L flux by momentum and solid angle. The simulations were validated against measurements from MIPP, NA61/SHINE, KOTO and E391a.
For K_L, the estimated number of decays inside the defined fiducial volumes was about 1.4×10^14 per year at 3 GeV, 9.1×10^15 at 6.75 GeV, 9.4×10^14 at 8 GeV and 3.3×10^15 at 120 GeV. The wide range reflects the different modeled beam scenarios and the way the proposed experiment defines its usable decay region.
The projections use 10^19 total kaons as a benchmark and express expected precision through two practical quantities: the total experimental efficiency and the number of background events. The main text estimates roughly 10^13 to 10^15 kaons per year in the defined fiducial volumes, leaving an estimated efficiency range of about 0.1 to 10^-3, depending on the background level.
What the proposed measurements could test
For the two golden modes together, the optimistic projection assumes uncertainties of 5% and 3%. A more conservative supplementary scenario uses 10% and 6%. These are modeled assumptions, not measured outcomes, and the projected precision is tied to the efficiency and background assumptions used in the analysis.
To estimate the CKM reach, the authors evaluate a log-likelihood over the rho-bar and eta-bar plane, using tree-level CKM inputs and profiling hadronic and short-distance nuisance parameters under Gaussian priors. They draw 1-sigma and 2-sigma confidence regions. In the resulting unitarity-triangle picture, the neutral mode constrains the triangle's height while the charged mode supplies a complementary ellipse, giving an independent cross-check of the CKM apex.
The neutral mode is aimed at a particularly demanding gap. Its current branching-ratio bound is below 2.2×10^-9 at a 90% confidence level, while the Standard Model prediction is about 3×10^-11. The neutral decay remains unobserved.
The study also translates an assumed 5% measurement into a projected new-physics scale reach. For the neutral golden mode, the reported reach is about 420 TeV under generic flavour violation and 5 TeV under minimal flavour violation. The corresponding entries for the charged mode are 216 TeV and 3.9 TeV. These figures are conditional projections for the stated measurement precision and flavour scenarios.
The practical obstacles are still unresolved
The kaon-yield estimate is given only within an order of magnitude. The authors note that a factor of 0.3 may be needed to account for beamline transmission, based on earlier studies. Exact instrumentation could also change the detector acceptance, the fraction of signal retained and the rejection of background events.
The neutral-kaon analysis faces a specific background concern: the beam contains more neutrons than K_L particles. The supplementary discussion considers angular cuts and an evacuated forward decay region as possible mitigations, but excludes a sophisticated background study because the beam and detector designs are not yet fixed.
The CKM forecasts are also conditional on their inputs. The calculation uses tree-level CKM information and Gaussian priors for hadronic and short-distance nuisance parameters, while the golden-mode scenarios assume particular measurement uncertainties. Its confidence regions therefore describe the consequences of those assumptions, rather than a completed experimental result.
The document is an arXiv version 1 preprint dated 27 August 2026. It presents a proposal and modeling study, not a completed measurement. Its acknowledgments name collaborators and discussants but do not state a funding source.
Paper data and sources
Original title: KOFFEE: Kaon Observables at Future Fermilab Experimental Extensions
Authors: Cari Cesarotti, Samuel Homiller
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-27
DOI: Not available
Original paper · Full text