On November 3 at CERN, physicists from the LHCb experiment announced that they had finally spotted a quarry hunted since 1984. It is not a new particle. It is a decay: the moment a short-lived particle called the B-zero meson falls apart into a muon and an antimuon. The Standard Model of particle physics says this happens roughly once in every ten billion B-zero decays. LHCb now has evidence that it happens, and at about the rate the theory predicts.

That might sound like a footnote. It is closer to a stress test of the most successful theory in physics, run at the edge of what any experiment can see.

The particle and the decay

A B-zero meson (physicists write it B0 or Bd) is a pair of quarks: a down quark bound to a beauty antiquark (its antiparticle partner, the anti-B-zero, has the charges reversed; experiments study the two together). Beauty quarks, also called bottom quarks, are heavier cousins of the up and down quarks that make up ordinary protons and neutrons. They are not part of ordinary, stable matter (cosmic rays make a few high in the atmosphere), but the Large Hadron Collider makes them in enormous numbers by smashing protons together. Once made, a beauty particle lives for about one and a half trillionths of a second before decaying.

Usually it decays into a messy spray of lighter particles. Very rarely, the beauty quark and the antiquark annihilate, and their energy reappears as a muon (a heavier relative of the electron) and its antiparticle. In the Standard Model this route is doubly suppressed. There is no direct, simple way for it to happen; it proceeds only through a complicated set of quantum interactions involving virtual heavy particles such as the W boson and the top quark. On top of that, a spinless meson can only turn into two muons if their spins line up in an awkward way, an effect called helicity suppression, which slashes the odds further. The result is a branching fraction, the share of B-zero mesons that decay this way, of about one in ten billion.

Why rarity is the point

Here is the counterintuitive part. Physicists care about this decay precisely because the Standard Model makes it so unlikely.

In quantum mechanics, a process that has no easy route takes every route available, including routes through particles too heavy to create directly. If nature contains undiscovered particles, such as additional Higgs bosons, new force carriers called Z-prime bosons, or exotic leptoquarks, they could open an extra channel for the decay and change how often it happens. When the expected rate is tiny, even a small new contribution shows up as a large relative change. That is why these decays have long been called "golden channels" in the search for new physics.

The University of Cambridge, home to several of the analysts, puts the scale in vivid terms. The LHC can directly probe structure down to about a hundred zeptometres, and the fact that it has not found new particles suggests they may hide at even shorter distances, as small as a zeptometre, a trillionth of a billionth of a metre. Precise measurements of rare decays let physicists feel the quantum influence of particles in that "zeptoverse" without ever producing them.

The hunt started with the CLEO experiment at Cornell, which set the first upper limit on the decay's rate in 1984. The more common sibling, the Bs meson (a beauty quark paired with a strange antiquark) decaying to two muons, fell first. LHCb reported first evidence for it in 2012; a joint LHCb and CMS analysis made the first observation in 2015; and LHCb alone observed it in 2017. That 2015 joint analysis also claimed evidence for the B-zero version, but later updates using data up to 2018 did not confirm it.

The B-zero decay is the harder target. According to LHCb, its Standard Model rate is about 30 times smaller than the Bs version. Even though the collider makes about four times as many B-zero mesons as Bs mesons, the expected B-zero signal is roughly seven times fainter than its sibling's. Worse, it sits right next to it in mass, buried under a mountain of ordinary collision debris.

What changed

Two things. First, hardware. In 2024 a major upgrade let LHCb record data at five times the rate it managed between 2010 and 2018, and improved its ability to tell different kinds of particles apart. Second, software. The team used machine-learning techniques to suppress the background that would otherwise bury the signal. "Extreme precision is required to pick the muons out of the overwhelming background," said Marco Santimaria of Italy's INFN Frascati laboratory, who co-led the 19-person analysis team with Harry Cliff of the University of Cambridge.

Using only data from 2024 and 2025, LHCb saw the B-zero signal at a significance of 3.0 standard deviations, or sigma. Combined with data from 2011 to 2018, it rose to 3.2 sigma.

What 3.2 sigma means

Particle physicists use sigma to express how surprising a signal would be if it were just a random fluctuation. Three sigma means that if only background were present, a fluctuation at least this large would show up only about 0.3% of the time. That is not the same as a 0.3% chance the signal is a fluke, but it is strong enough to count as evidence. That is the conventional threshold for "evidence." It is not a discovery. Discovery requires five sigma, a far stricter bar, because physicists search through so much data that three-sigma flukes do occasionally appear and later vanish. The B-zero decay itself is a cautionary example: the 2015 evidence did not hold up.

In short: for the first time, a single experiment has produced a signal too strong to dismiss.

The numbers

According to the collaboration's preprint, the new data alone give a branching fraction of about 1.26 in ten billion, with a statistical uncertainty of roughly 0.5 in ten billion. Combined with older data, the figure is 1.14 in ten billion. The Standard Model predicts about 1.03 in ten billion. Within the uncertainties, theory and experiment agree.

LHCb also produced the most precise measurements yet of the Bs sibling: a branching fraction of 3.31 in a billion, against a predicted 3.66, and an "effective lifetime" of 1.60 picoseconds, plus or minus 0.16, in agreement with the Standard Model. That lifetime matters because some new-physics models would leave the decay rates untouched while changing how the decay depends on time. LHCb also tracks the ratio of the two branching fractions, which tests new-physics models in a different way from either rate alone.

So, no new physics?

Not this time, and that is still useful. Every measurement that agrees with the Standard Model shrinks the room in which new theories can live. LHCb says the results "isolate even more precisely the parameter region in which these new models can exist," which tells theorists where not to look and experimenters where to look harder.

There is also an honest caveat. The paper, posted to arXiv as 2611.00230 and submitted to the journal Physical Review X, has not yet completed peer review, and the uncertainties on the B-zero rate are still large. A deviation from the Standard Model could still be hiding inside them.

What comes next

More data. The High-Luminosity LHC is due to start in 2030, and a further LHCb upgrade planned for the middle of the 2030s could raise the data-collection rate by another factor of ten. With that, the B-zero decay should move from evidence to a five-sigma observation and then to a precision measurement, where a small disagreement with theory would be impossible to wave away.

"We're knocking on the door of the zeptoverse now," Cliff said. "I can't wait to see what we find as we explore deeper."

Forty-two years after the hunt began, physicists have caught a one-in-ten-billion decay, and found it behaving exactly as the Standard Model predicted. The theory has survived another test. The question is how many more it can pass.