A particle that almost never decays in a particular way has left a trace in one experiment. Researchers with LHCb at CERN report evidence for B⁰ → μ⁺μ⁻, in which a neutral beauty meson produces a positively charged muon and a negatively charged muon. The collaboration announced the result at a CERN seminar on 3 November. Its paper was submitted to Physical Review X and posted on arXiv; Phys.org reported the result on 5 November. The milestone follows more than four decades of searching, but its statistical strength remains below the threshold for a discovery claim.
A signal after four decades
The LHCb collaboration reports the first evidence for B⁰ → μ⁺μ⁻ from a single experiment. The signal appears in proton-proton collision data collected in 2024 and 2025 during the Large Hadron Collider's Run 3. Using that new sample alone, the analysis reaches a statistical significance of 3.0 standard deviations. Combining it with LHCb measurements from the first two runs raises the figure to 3.2. These are distinct results: 3.2 is the combined significance, not the strength of the new data alone.
The search began in 1984, when Cornell's CLEO experiment set an upper limit rather than observing the decay. A joint LHCb-CMS analysis reported a hint in 2015, but later analyses did not confirm it. The new result is therefore the first single-experiment evidence, not the first historical suggestion of a signal. That distinction matters: rare-event claims can strengthen or weaken as larger samples test earlier excesses. The current paper uses the upgraded LHCb detector and combines its new data with the experiment's earlier measurements.
Sources: Evidence for the decay B^{0}→μ^{+}μ^{–} and improved measurements of B_{s}^{0}→μ^{+}μ^{–} · Evidence for the decay B⁰ → μ⁺μ⁻ and improved measurements of the decay B⁰ₛ → μ⁺μ⁻ · LHCb finds evidence of one-in-ten-billion particle decay · LHCb identifies a decay that occurs once in ten billion
What a B⁰ meson is
A B⁰ meson is a short-lived, electrically neutral particle made from two quark constituents: a beauty, or bottom, antiquark and a down quark. Quarks are elementary building blocks that combine into particles such as protons and mesons. In this decay, the beauty antiquark and down quark annihilate, producing two muons with opposite electric charges. Muons resemble electrons but are much heavier; they belong to the lepton family. The notation μ⁺μ⁻ simply identifies the positive and negative members of the pair.
The analysis also measures the related B⁰ₛ → μ⁺μ⁻ decay. The B⁰ₛ contains a beauty antiquark paired with a strange quark, and its muon-pair decay is less rare and had already been observed. In the combined LHCb data, its branching fraction is about 3.25 per billion B⁰ₛ decays. The B⁰ channel is much harder: the measured rate is about 1.14 per ten billion B⁰ decays. A branching fraction is the probability of one specified decay route, not the share of all proton collisions that produce muons.
Sources: Evidence for the decay B⁰ → μ⁺μ⁻ and improved measurements of the decay B⁰ₛ → μ⁺μ⁻ · LHCb finds evidence of one-in-ten-billion particle decay · LHCb identifies a decay that occurs once in ten billion · After 40 years of searching, physicists find evidence for an ultrarare particle decay
Why the decay is extraordinarily rare
The rarity is not simply because the B⁰ meson is short-lived. The transition changes quark flavour without changing electric charge, a process called a flavour-changing neutral current. In the Standard Model it cannot occur through the simplest direct interaction. Instead, it proceeds through higher-order quantum processes involving virtual heavy particles, including the W boson and top quark. These loop processes are strongly suppressed, so only a tiny fraction of B⁰ mesons are expected to produce a muon pair.
A second suppression is linked to spin and the weak interaction's preference for particular handedness states. The initial meson has zero total angular momentum, restricting the allowed two-muon configuration. This makes the decay extremely uncommon but scientifically useful. An unknown heavy particle could contribute virtually to the same quantum process and subtly alter its rate, even if the collider could not produce it directly. Physicists therefore compare the measured probability with the Standard Model prediction. Rarity creates sensitivity; it is not itself evidence of new physics.
Sources: Evidence for the decay B⁰ → μ⁺μ⁻ and improved measurements of the decay B⁰ₛ → μ⁺μ⁻ · LHCb finds evidence of one-in-ten-billion particle decay · LHCb identifies a decay that occurs once in ten billion
How LHCb separated signal from noise
The analysis uses 18 inverse femtobarns of Run 3 proton-proton data recorded at 13.6 teraelectronvolts. LHCb's 2022 Upgrade I detector can operate at about five times the collision rate of its predecessor. INFN reports that the experiment analyses more than 100 million proton-proton collisions each second, generating a data flow of roughly four terabytes per second. The sought-after decay is a tiny signal buried among an enormous number of other interactions and particle decays.
The team had to identify genuine muons and reject ordinary processes that imitate two oppositely charged tracks. The paper describes a software trigger, particle-identification requirements and a boosted decision-tree classifier to rank candidates by how signal-like they appear. Researchers then fitted the invariant mass of the two-muon system to estimate the signal above background. To reduce the risk of tuning the analysis toward a desired answer, the mass window containing the candidate signals was kept hidden until the selection methods were finalised.
Sources: Evidence for the decay B⁰ → μ⁺μ⁻ and improved measurements of the decay B⁰ₛ → μ⁺μ⁻ · LHCb finds evidence of one-in-ten-billion particle decay · LHCb identifies a decay that occurs once in ten billion
Three sigma is evidence, not discovery
Statistical significance measures how unusual an observed excess would be if the background-only explanation were correct. In particle physics, about three standard deviations is commonly described as evidence; five is the widely used threshold for a discovery claim. LHCb's combined 3.2-sigma result clears the evidence level but remains below that standard. It does not mean there is a 99.9 per cent probability that the decay exists, or that the Standard Model is wrong. It describes how compatible the excess is with a specified background-only hypothesis.
The distinction is substantive. A signal at this level can still be affected by statistical fluctuation, and the measured B⁰ branching fraction has a large uncertainty. The collaboration's paper was submitted to Physical Review X and posted as an arXiv preprint; as of 7 November, the sources reviewed for this report do not show a completed peer-reviewed publication. The result is an experimental milestone, but remains open to scrutiny and confirmation. More data could strengthen the case, reduce uncertainty or show that the excess was less stable than it first appeared.
Sources: Evidence for the decay B^{0}→μ^{+}μ^{–} and improved measurements of B_{s}^{0}→μ^{+}μ^{–} · Evidence for the decay B⁰ → μ⁺μ⁻ and improved measurements of the decay B⁰ₛ → μ⁺μ⁻ · LHCb finds evidence of one-in-ten-billion particle decay · After 40 years of searching, physicists find evidence for an ultrarare particle decay
A sharper test of the Standard Model
Combining all three LHCb data periods gives a B⁰ → μ⁺μ⁻ branching fraction of 1.14 × 10⁻¹⁰. Its statistical uncertainty is +0.42 and −0.39, with an additional systematic uncertainty of +0.06 and −0.05, in the same units. The Standard Model prediction quoted in the paper is (1.03 ± 0.05) × 10⁻¹⁰. The measured central value is slightly higher, but the uncertainty is substantial and the prediction lies within the measured range. LHCb therefore reports agreement with the Standard Model, not a significant discrepancy.
The companion B⁰ₛ result is also consistent with its prediction and is the world's most precise single-experiment measurement of that branching fraction. The new B⁰ evidence narrows the room available to some theories that add particles or interactions, but it does not identify such particles or establish new physics. The next step is more data and a stronger statistical case, ideally reaching five standard deviations, alongside tighter measurements of both decay rates and the B⁰ₛ effective lifetime. The achievement is that a detector handling a torrent of collisions has made a process occurring roughly once in ten billion decays testable.
Sources: Evidence for the decay B⁰ → μ⁺μ⁻ and improved measurements of the decay B⁰ₛ → μ⁺μ⁻ · LHCb finds evidence of one-in-ten-billion particle decay · LHCb identifies a decay that occurs once in ten billion · After 40 years of searching, physicists find evidence for an ultrarare particle decay
