Mixing and CP violation
The mixing of the three neutrino flavours is described by the PMNS matrix, with three angles that are now measured to a few percent and one phase that is still poorly known. If that phase violates CP symmetry, neutrinos and antineutrinos behave differently, which may be connected to why the universe is made of matter.
Try it yourself: oscillation calculator →
Beginner
Three flavours, three masses
There are three kinds of neutrino, called flavours: the electron neutrino, the muon neutrino and the tau neutrino. Each is named after the charged particle it appears with when it is created. But a neutrino of a given flavour does not have a single, definite mass. Instead, it is a blend of three mass states with slightly different masses, much as a musical chord is a blend of three notes. Because the three notes travel at slightly different rhythms, the blend shifts as the neutrino flies, and the flavour you detect at the end may differ from the one you started with. This is neutrino oscillation.
The mixing recipe
The recipe that says how much of each mass state goes into each flavour is a table of numbers called the PMNS matrix, named after Bruno Pontecorvo, Ziro Maki, Masami Nakagawa and Shoichi Sakata. The whole table can be described by three angles and one extra number called a phase. You can picture the angles as the settings of three turning knobs. Two of the knobs are turned a long way, around 33 and 45 degrees, and one is turned only a little, about 8.5 degrees. Experiments have measured all three angles quite precisely3.
This is very different from quarks, the particles inside protons and neutrons. Quarks also mix, but only slightly. Neutrinos mix strongly, and nobody yet knows why.
The missing phase
The fourth number, the phase, is called δ (delta). It is still poorly known. It matters because it decides whether neutrinos and antineutrinos oscillate in exactly the same way. Physicists call this a test of CP symmetry, the idea that the laws of nature stay the same if you swap every particle for its antiparticle and look at the world in a mirror. If δ has the right value, muon neutrinos would turn into electron neutrinos more often than muon antineutrinos turn into electron antineutrinos, or the other way round.
How it is measured
Large particle accelerators produce beams of muon neutrinos and, by flipping a magnetic switch, beams of muon antineutrinos. The beams are sent hundreds of kilometres through the Earth to a huge detector. The Japanese experiment T2K fires its beam 295 kilometres across Japan, and the American experiment NOvA 810 kilometres from Chicago to Minnesota. In 2025 the two teams combined their data for the first time5. The results hint that CP symmetry may be broken for neutrinos, but they are not yet conclusive.
Two giant new detectors are being built to settle the question: DUNE in the United States, which will send neutrinos 1,300 kilometres underground, and Hyper-Kamiokande in Japan, a water tank many times larger than its predecessor.
Why it matters
Right after the Big Bang, matter and antimatter should have been made in equal amounts and destroyed each other almost completely. Yet a small surplus of matter survived, and everything we see is made from it. Explaining this surplus requires processes that treat matter and antimatter differently. One popular idea, called leptogenesis, suggests that very heavy relatives of neutrinos produced this imbalance in the early universe6. Finding CP violation in ordinary neutrinos would not prove this idea, but it would show that neutrinos can tell matter and antimatter apart.
Quiz
Test your knowledge
4 questions on this level · Print as worksheet
Intermediate
From two flavours to three
In 1962 Ziro Maki, Masami Nakagawa and Shoichi Sakata proposed that the electron and muon neutrinos could be mixtures of two other states1, and Bruno Pontecorvo developed the idea of flavour oscillation. With three flavours the mixing is described by a 3 × 3 unitary matrix, the PMNS matrix. It contains three mixing angles, θ₁₂, θ₂₃ and θ₁₃, and one CP-violating phase δ. If neutrinos are Majorana particles, two further phases appear, but they have no effect on oscillations.
Who measured what
- θ₁₂ ≈ 33.7°: solar neutrinos (Homestake, SAGE, GALLEX, Super-Kamiokande, SNO, Borexino) and the reactor experiment KamLAND; now also JUNO.
- θ₂₃ ≈ 43–49°: atmospheric neutrinos at Super-Kamiokande (1998) and IceCube–DeepCore, and the accelerator experiments K2K, MINOS, T2K and NOvA. Whether θ₂₃ is slightly below or above 45° (the octant problem) is still open.
- θ₁₃ ≈ 8.5°: measured in 2012 by the reactor experiments Daya Bay and RENO, with earlier hints from T2K and Double Chooz.
The global analysis NuFIT 6.0 (2024) gives these angles to between about 1.5 % and a few percent3. Compared with the quark mixing matrix, whose largest angle, the Cabibbo angle, is about 13°, neutrino mixing is strikingly large.
CP violation and the Jarlskog invariant
CP violation means that a process and its mirror-image antimatter version proceed at different rates. In oscillations it shows up as a difference between P(νμ → νe) and P(ν̄μ → ν̄e). Its size is controlled by a single combination of the mixing parameters, introduced by Cecilia Jarlskog in 1985 for quarks2 and called the Jarlskog invariant.
For quarks J is about 3 × 10⁻⁵. Because the neutrino angles are large, leptonic CP violation could be up to about a thousand times stronger, if δ is far from 0 and 180°.
Long-baseline experiments
T2K sends a beam from J-PARC on Japan's east coast 295 km to Super-Kamiokande, using an off-axis beam that peaks near 0.6 GeV. NOvA sends the Fermilab NuMI beam 810 km to a 14 kt detector in northern Minnesota, at about 2 GeV. Both alternate between neutrino and antineutrino modes by reversing the horn polarity. In 2020 T2K excluded almost half of the possible values of δ at 3σ, with a preference for δ near 270°4. NOvA's data alone prefer values closer to 180° in the normal ordering. The first joint fit, published in October 2025, measured the atmospheric mass splitting to better than 2 %, found no strong preference for either mass ordering, and showed that if the ordering is inverted, CP conservation is disfavoured at the level of evidence5.
The next generation
DUNE will send a wide-band beam from Fermilab 1,300 km to liquid-argon detectors at the Sanford Underground Research Facility in South Dakota; the long baseline gives strong matter effects and resolves the mass ordering along with δ7. Hyper-Kamiokande, a 260,000-tonne water Cherenkov detector, will use an upgraded J-PARC beam on the T2K baseline from about 20288. Each aims to discover CP violation at 5σ for a large fraction of possible δ values within about ten years of running.
Link to matter–antimatter asymmetry
In leptogenesis, proposed by Masataka Fukugita and Tsutomu Yanagida in 1986, heavy right-handed neutrinos decay unequally into leptons and antileptons, and processes in the early universe convert part of this lepton excess into the baryon excess we observe6. Observing CP violation in light-neutrino oscillations would make such a mechanism more plausible but not prove it, because the CP phases relevant for the heavy neutrinos are in general not the same as δ.
Quiz
Test your knowledge
4 questions on this level · Print as worksheet
Expert
Parameterisation
The flavour states are να = Σᵢ U*_αi νᵢ. In the standard (PDG) parameterisation the PMNS matrix is the product of three rotations, the middle one carrying the Dirac phase, times a diagonal Majorana-phase matrix:
Unitarity of a 3 × 3 matrix leaves 3 angles and 1 phase after rephasing the charged-lepton and (for Dirac) neutrino fields. Direct tests of unitarity, by measuring the rows and columns independently, constrain deviations at the percent level in the e-row and much more weakly in the μ- and τ-rows; non-unitarity would signal heavy neutral leptons or other new states.
Current values
NuFIT 6.0 (data as of 2024) gives for normal ordering, without (IC19) and with (IC24) the tabulated Super-Kamiokande atmospheric χ²3:
- sin²θ₁₂ = 0.307 (+0.012, −0.011), θ₁₂ = 33.68° (+0.73, −0.70); insensitive to the atmospheric data.
- sin²θ₂₃ = 0.561 (+0.012, −0.015) [IC19] or 0.470 (+0.017, −0.013) [IC24]; the octant flips depending on the data set, and the 3σ range 0.430–0.596 spans both.
- sin²θ₁₃ = 0.02195 (+0.00054, −0.00058), θ₁₃ = 8.52° ± 0.11° [IC19].
- δ = 177° (+19, −20) [IC19] or 212° (+26, −41) [IC24]; for inverted ordering 285° or 274°.
- Δm²₂₁ = (7.49 ± 0.19) × 10⁻⁵ eV²; Δm²₃ℓ = +2.534 (+0.025, −0.023) × 10⁻³ eV² [IC19] or +2.513 (+0.021, −0.019) × 10⁻³ eV² [IC24].
For normal ordering the global fit is consistent with CP conservation within about 1σ; for inverted ordering, values near δ = 270° are favoured over CP conservation at more than 3σ. The maximal Jarlskog invariant is Jmax = 0.0333 ± 0.00073, compared with JCKM ≈ 3 × 10⁻⁵ for quarks2.
How δ appears in oscillations
In vacuum the CP asymmetry is governed by J and all three oscillation phases Δᵢⱼ = Δm²ᵢⱼ L / 4E:
At the first oscillation maximum the δ-dependent term in P(νμ → νe) is roughly proportional to sin 2θ₁₂ sin 2θ₂₃ sin 2θ₁₃ (Δm²₂₁/Δm²₃₁) and changes the appearance probability by tens of percent. Matter effects add a fake CP asymmetry of the same sign structure as the mass ordering: for normal ordering they enhance neutrino and suppress antineutrino appearance, growing roughly linearly with energy and baseline. At T2K (295 km, 0.6 GeV) the matter-induced asymmetry is of order 10 %, at NOvA (810 km, 2 GeV) roughly three times larger; NOvA therefore has better ordering sensitivity but stronger ordering–δ degeneracy.
T2K and NOvA
T2K's 2020 analysis gave 3σ intervals for δ of [−3.41, −0.03] rad (normal ordering) and [−2.54, −0.32] rad (inverted), disfavouring CP conservation at about 95 % confidence4. NOvA's appearance data, by contrast, do not show the large νe enhancement expected for δ ≈ 270° in normal ordering, and the two experiments' preferred regions overlap only partly in that ordering. The 2025 joint fit, accounting for correlated cross-section systematics, yields Δm²₃₂ = 2.43 (+0.04, −0.03) × 10⁻³ eV² (normal) and 2.48 (+0.03, −0.04) × 10⁻³ eV² (inverted), 3σ δ intervals of [−1.38π, 0.30π] (normal) and [−0.92π, −0.04π] (inverted), and no strong ordering preference5.
DUNE and Hyper-Kamiokande
DUNE combines the 1,300 km baseline from Fermilab to SURF with a wide-band beam of 1.2 MW (upgradable to 2.4 MW) and four liquid-argon TPC far-detector modules of about 17 kt each, so that the first and second oscillation maxima are both sampled7. Its design goal is a 5σ mass-ordering determination for all δ and 5σ CP-violation discovery for half of the δ range after about ten years of staged running. Hyper-Kamiokande, with 188 kt fiducial water mass, an upgraded 1.3 MW J-PARC beam and new near and intermediate detectors, projects 5σ CP-violation sensitivity for about 57 % of δ values in ten years8. The two approaches differ in their systematic profiles: Hyper-K is statistics-rich at low energy where quasi-elastic scattering dominates, DUNE exploits spectral information across 0.5–5 GeV where resonance and deep-inelastic processes matter. Cross-section uncertainties, especially νe/νμ and ν/ν̄ differences in nuclear targets, are the leading systematics for both.
Leptogenesis
The observed baryon-to-photon ratio, ηB ≈ 6 × 10⁻¹⁰, requires the Sakharov conditions. In thermal leptogenesis6, CP-violating, out-of-equilibrium decays of the lightest heavy Majorana neutrino N₁ create an asymmetry in B − L, and electroweak sphalerons convert a fraction of about one third of it into baryon number. In hierarchical type-I see-saw models this requires M₁ ≳ 10⁹ GeV (the Davidson–Ibarra bound). The high-energy CP phases are related to δ and the Majorana phases only through the unknown Yukawa matrix, so no model-independent prediction connects them. In specific scenarios, for example when flavour effects dominate, low-energy δ alone can suffice, which gives the measurement indirect cosmological weight.
Quiz
Test your knowledge
4 questions on this level · Print as worksheet
Open questions
- The value of δ and whether leptonic CP symmetry is violated; DUNE and Hyper-Kamiokande are designed to answer this in the 2030s.
- The octant of θ₂₃: below or above 45°, with current fits flipping between data sets.
- Why lepton mixing is large while quark mixing is small, and whether the 3 × 3 PMNS matrix is exactly unitary.
Terms to explore
- Neutrino mixing
- PMNS matrix
- Mixing angle θ12
- Mixing angle θ23
- Mixing angle θ13
- CP-Violating phase δCP
- Jarlskog invariant
- Leptonic CP violation
- Quark vs lepton mixing
- θ23 octant
- CP-Violation sensitivity (δ_CP)
- Leptogenesis
People
- Ziro Maki
- Masami Nakagawa
- Shoichi Sakata
- Bruno Pontecorvo
- Nicola Cabibbo
- Masataka Fukugita
- Tsutomu Yanagida
- Kōichirō Nishikawa
On the timeline
- 1957Pontecorvo proposes neutrino oscillations
- 1962Maki, Nakagawa and Sakata introduce neutrino mixing
- 1986Fukugita and Yanagida propose leptogenesis
- 2011T2K finds hints of electron-neutrino appearance
- 2012Daya Bay discovers a non-zero θ₁₃
- 2012RENO confirms θ₁₃
- 2016NOvA reports its first oscillation results
- 2020T2K constrains CP violation in neutrinos
- 2025NOvA and T2K publish a joint analysis
- 2020Hyper-Kamiokande construction is approved
- 2017Groundbreaking for LBNF and DUNE
References
- Z. Maki, M. Nakagawa, S. Sakata, Remarks on the unified model of elementary particles, Prog. Theor. Phys. 28, 870 (1962), doi:10.1143/PTP.28.870
- C. Jarlskog, Commutator of the quark mass matrices in the standard electroweak model and a measure of maximal CP nonconservation, Phys. Rev. Lett. 55, 1039 (1985), doi:10.1103/PhysRevLett.55.1039
- I. Esteban, M. C. Gonzalez-Garcia, M. Maltoni, T. Schwetz, A. Zhou, NuFit-6.0: updated global analysis of three-flavor neutrino oscillations, JHEP 12, 216 (2024), doi:10.1007/JHEP12(2024)216
- K. Abe et al. (T2K Collaboration), Constraint on the matter–antimatter symmetry-violating phase in neutrino oscillations, Nature 580, 339 (2020), doi:10.1038/s41586-020-2177-0
- NOvA and T2K Collaborations, Joint neutrino oscillation analysis from the T2K and NOvA experiments, Nature 646, 818 (2025), doi:10.1038/s41586-025-09599-3
- M. Fukugita, T. Yanagida, Baryogenesis without grand unification, Phys. Lett. B 174, 45 (1986), doi:10.1016/0370-2693(86)91126-3
- B. Abi et al. (DUNE Collaboration), Deep Underground Neutrino Experiment (DUNE) Far Detector Technical Design Report, Volume II: DUNE Physics, arXiv:2002.03005 (2020), arXiv:2002.03005
- K. Abe et al. (Hyper-Kamiokande Proto-Collaboration), Hyper-Kamiokande Design Report, arXiv:1805.04163 (2018), arXiv:1805.04163
Last updated: · Editorial standard
Cite this page
Text
Neutrino Atlas (2026). “Mixing and CP violation”. Retrieved 22 November 2026, from https://neutrino-atlas.com/learn/mixing-and-cp-violation/BibTeX
@misc{neutrinoatlas2026earnmixingandcpviolation,
title = {Mixing and CP violation},
author = {{Neutrino Atlas}},
year = {2026},
url = {https://neutrino-atlas.com/learn/mixing-and-cp-violation/},
note = {Accessed 2026-11-22},
language = {english}
}