B⁰ decay to two muons: LHCb reports 3.2-sigma evidence, short of discovery

LHCb’s combined data give 3.2 standard deviations for a B⁰ meson decaying to two muons, a rate of about 1.14 in 10 billion decays that matches the Standard Model, but short of the 5-sigma bar for discovery.

Preprint, not yet peer reviewed · Particle-collider measurement · Evidence, not discovery

The LHCb collaboration at CERN reports the first evidence from a single experiment that a neutral B⁰ meson can decay into a muon and an antimuon. The signal reaches 3.0 standard deviations in the 2024–2025 data and 3.2 once earlier runs are added. The measured rate agrees with the Standard Model’s prediction, and the team also reports its most precise values yet for the sister decay of the Bs⁰ meson. The catch: 3.2 sigma is below the 5-sigma threshold physicists use to claim a discovery, and the paper has not yet been peer reviewed.

Evidence at a glance

Is this a big deal?

A solid step, not a breakthrough. It is the first time one experiment has crossed the 3-sigma line for this decay, and the Bs⁰ numbers are the most precise on record. Nothing in the data contradicts the Standard Model.

What scientists found

A B⁰ meson is a short-lived particle made of a beauty (b) antiquark and a down quark. The Standard Model predicts it turns into two muons only about once in every 10 billion decays, a branching fraction of (1.03 ± 0.05) × 10⁻¹⁰. LHCb measures (1.14 +0.42/−0.39 +0.06/−0.05) × 10⁻¹⁰ from all data combined, where the first uncertainty is statistical and the second systematic.

That is about 11% above the prediction, but the uncertainty is about 37%, so the two agree. By SciQuest’s simple estimate, treating the errors as symmetric and Gaussian, the gap is roughly 0.3 standard deviations.

The 3.2σ figure answers a different question, the chance that background alone produced an excess this large. A one-sided 3.2σ excess occurs by chance about once in 1,450 trials (SciQuest calculation). It does not measure agreement with theory.

Run 3 data alone give B(B⁰ → μ⁺μ⁻) = (1.26 +0.51/−0.47 +0.06/−0.05) × 10⁻¹⁰ at 3.0σ, and a Bs⁰ effective lifetime of 1.42 ± 0.18 ± 0.02 ps. For the Bs⁰ branching fraction, the combined value sits about 1.3 standard deviations below the prediction by the same rough estimate, which is not a significant shortfall. The paper reports all results as consistent with the Standard Model.

How they found it

LHCb is a detector at CERN’s Large Hadron Collider built to study particles containing beauty quarks. Its upgraded version, which recorded the 2024–2025 data used here, can run at five times the previous collision rate and replaced its first-level hardware trigger with software. The 2024–2025 sample amounts to 18 inverse femtobarns (fb⁻¹, a measure of how many collisions were recorded), twice the 9 fb⁻¹ from 2011–2018.

The search is for a needle in a very large pile: two oppositely charged muons that together carry the mass of a B meson. The team trained a boosted decision tree, a machine-learning classifier, to reject random muon pairs, then fitted the dimuon mass distribution in five classifier bins. Signal counts became branching fractions by comparison with two common decays, B⁰ → K⁺π⁻ and B⁺ → J/ψ K⁺.

To avoid steering toward a wanted answer, the mass window from 5200 to 5450 MeV/c², where the signals sit, was not examined until every method was fixed. Control decays such as Bs⁰ → K⁺K⁻ checked the lifetime measurement; their results landed within 0.012 ps of the known values, which became a systematic uncertainty.

Why it matters to you

Nothing about this result changes daily life, and it is not a health or safety finding. Its value is as a test of the Standard Model, the theory behind subatomic particles and how they interact, and the Higgs boson completed its list of particles. The theory cannot explain dark matter or why the universe holds more matter than antimatter, so physicists look for places where it might crack.

Rare decays are such places. Because the Standard Model allows this decay only through delicate quantum loops, unseen heavy particles could add their own contribution and change the rate. The paper says these measurements constrain new physics at energy scales beyond what the LHC can produce directly.

What the evidence does not show

It does not show that the B⁰ decay is real beyond reasonable doubt. At 3.2σ, a fluctuation remains possible, and this decay has been here before: a joint LHCb and CMS analysis reported evidence for it in 2015, and later analyses did not confirm it. The current experimental upper limit of 1.6 × 10⁻¹⁰ (at 90% confidence) came from a 2020 combination of ATLAS, CMS and LHCb. The new value, 1.14 × 10⁻¹⁰, sits below that limit, so the two do not conflict.

It does not show new physics. Both branching fractions and the ratio agree with the Standard Model within their uncertainties. The Bs⁰ lifetime is dominated by statistics: its recast parameter has an uncertainty (±1.92) larger than the whole physical range of −1 to +1.

The result is also a preprint. Peer review at Physical Review X may change details, though the collaboration’s own cross-checks are extensive.

How this fits what we already knew

The search began in 1984, when the CLEO experiment set an upper limit of 2.0 × 10⁻⁴ on the B⁰ decay, about two million times the predicted rate (SciQuest calculation: 2.0 × 10⁻⁴ ÷ 1.03 × 10⁻¹⁰ ≈ 1.9 million). The Bs⁰ decay was observed by LHCb and CMS together in 2015 and by LHCb alone in 2017. The new work supersedes LHCb’s earlier Run 1 and 2 results.

The Standard Model predicts the B⁰ decay is about 35 times rarer than the Bs⁰ decay, mostly because of the strengths of the quark couplings involved (3.66 × 10⁻⁹ ÷ 1.03 × 10⁻¹⁰ ≈ 35.5 by SciQuest’s arithmetic). The measured ratio, 0.035, is larger than the predicted 0.028, but with an uncertainty of roughly 40% it is no tension.

Independent perspective

SciQuest did not interview an independent physicist, and the quotes in the Birmingham and Manchester releases come mainly from LHCb members, so they are not independent. The paper’s own context is a useful check: it notes that several deviations from the Standard Model seen in related b-quark decays over the past decade are hard to interpret because of difficult quark-and-gluon calculations, whereas these two-muon decays are theoretically cleaner. That is the main reason physicists care about them.

How other headlines framed it

• LHCb outreach stuck to the measurement itself: first single-experiment evidence, 3σ rising to 3.2σ, and agreement with the Standard Model.
• University of Birmingham stressed UK leadership and a 42-year search, and said the results match the Standard Model.
• University of Manchester (also carried by Phys.org) framed it as a step toward the “Zeptoverse,” the unexplored scale of about 10⁻²¹ metres, while noting that 5σ was not reached.
• Yahoo Noticias (in Spanish) called it a promising but not conclusive result, and explained that the paper is an unreviewed preprint and that the 2015 hint was not confirmed.

No outlet exaggerated the result, though the “step closer” framing invites more optimism than the data support.

What happens next

More data are the main route. LHCb continues to record collisions, and the outreach post and press coverage point to a further detector upgrade later in the 2030s. Whether the B⁰ signal grows toward 5σ, or shrinks like its 2015 predecessor, depends on that data. The ATLAS and CMS experiments can also test it independently.

In one sentence

LHCb has the first single-experiment evidence, at 3.2σ, that B⁰ mesons decay to two muons at roughly the rate the Standard Model predicts, and only more data can turn that into a discovery.

Read next

• Why the Higgs boson is called the “God particle”: the last piece of the Standard Model to be found.
• Jet quenching seen in RHIC oxygen collisions: another collider result with its limits spelled out.
• The parts of an atom: the subatomic building blocks behind the particles in this story.

Sources and disclosure

1. LHCb collaboration. “Evidence for the decay B⁰ → μ⁺μ⁻ and improved measurements of the decay Bs⁰ → μ⁺μ⁻.” Preprint, arXiv:2611.00230 (CERN-EP-2026-287, LHCb-PAPER-2026-053), 30 October 2026; submitted to Physical Review X.
2. University of Birmingham, “Physicists uncover evidence of one of nature’s rarest particle decays”, 3 November 2026.
3. LHCb outreach, “Evidence for the decay B0→μ+μ– and improved measurements of Bs0→μ+μ–”, 3 November 2026.
4. Standard Model predictions as cited in the paper: M. Beneke, C. Bobeth and R. Szafron, “Power-enhanced leading-logarithmic QED corrections to B_q → μ⁺μ⁻,” JHEP 10 (2019) 232, arXiv:1908.07011 (SciQuest read the numbers in the LHCb paper, not this reference).
5. Coverage read for framing: University of Manchester (3 November 2026), Phys.org (3 November 2026), Yahoo Noticias (4 November 2026).

Disclosure: This article was drafted with AI assistance and checked against the sources above, including the full text of the preprint. Calculations by SciQuest: measured ÷ predicted ratios (1.14 ÷ 1.03 = 1.11; 3.25 ÷ 3.66 = 0.89; 0.035 ÷ 0.0280 = 1.25); rough deviations from the predictions, 0.11 ÷ 0.42 ≈ 0.3σ and 0.41 ÷ √(0.29² + 0.14²) ≈ 1.3σ, which treat errors as symmetric and Gaussian and are not the paper’s numbers; one-sided Gaussian tail probability for 3.0σ (1 in about 740), 3.2σ (1 in about 1,450) and 5σ (1 in about 3.5 million). The graphic is a SciQuest original built from the paper’s reported values. Spotted an error? See our corrections policy.