Daya Bay and the Discovery of θ₁₃

In March 2012, the Daya Bay collaboration announced a measurement of the last unknown neutrino mixing angle, ${\theta }_{13}$, at five standard deviations significance. The result was simultaneously expected and surprising: expected because prior experiments had set progressively tighter bounds that made a non-zero value increasingly likely; surprising because the central value — ${\sin }^{2}2{\theta }_{13}\approx 0.092$ — was large enough that long-baseline CP-violation searches instantly became feasible on a timescale of a decade rather than two.

Before Daya Bay, the three-flavor oscillation framework had been pieced together across two decades of disparate measurements. Solar experiments had determined ${\theta }_{12}$. Atmospheric and accelerator experiments had determined ${\theta }_{23}$. The third angle, ${\theta }_{13}$, remained stubbornly below the sensitivity of every previous attempt. Daya Bay’s result completed the three-angle picture and opened the door to the next frontier: measuring CP violation in the lepton sector.

This is the story of the experiment that closed one chapter of neutrino physics and opened the next.

Why θ₁₃ mattered

By 2000, the picture looked like this. Solar neutrinos oscillated with a large mixing angle (${\theta }_{12}\approx 34°$) and a small squared-mass splitting ($\Delta m_{21}^2\approx 7.4\times 10^{-5}$ eV²). Atmospheric neutrinos oscillated with a near-maximal angle (${\theta }_{23}\approx 45°$) and a larger splitting ($|\Delta m_{31}^2|\approx 2.5\times 10^{-3}$ eV²). The third angle ${\theta }_{13}$, connecting the solar and atmospheric sectors, was bounded only from above.

The CHOOZ reactor experiment in France had published the strongest bound in 1999: ${\sin }^{2}2{\theta }_{13}<0.17$ at 90% confidence. The experiment searched for ${\bar{\nu }}_e$ disappearance at a baseline of 1 km from the Chooz reactor complex, saw no deficit, and set the limit. This ruled out the tri-bimaximal predictions of several flavor models and kept open the question of whether ${\theta }_{13}$ was merely small or actually zero.

The practical importance of ${\theta }_{13}$ was immense. The CP-violating observable in long-baseline oscillations is $\Delta P=P({\nu }_{\mu }\to {\nu }_e)-P({\bar{\nu }}_{\mu }\to {\bar{\nu }}_e)\propto \sin (2{\theta }_{13})\sin ({\delta }_{CP})\cdot \text{(other factors)}$ If $\sin (2{\theta }_{13})=0$, no CP asymmetry can ever be observed — regardless of the actual Dirac phase. The feasibility of future experiments like NOvA, T2K, and DUNE depended entirely on ${\theta }_{13}$ being non-zero and, ideally, not too small.

Three reactor experiments were designed and built specifically to resolve this question: Daya Bay in China, Double Chooz in France, and RENO in South Korea.

The Daya Bay site

Daya Bay is a nuclear power complex in Guangdong province, southern China, about 55 km northeast of Hong Kong. In 2012 it comprised three twin-core pressurised-water reactors (Daya Bay 1/2, Ling Ao 1/2, Ling Ao II 1/2), producing a combined thermal power of 17.4 GW — one of the most intense antineutrino sources on Earth. Each reactor emits approximately $10^{20}$ antineutrinos per second from the decay of fission fragments.

The experiment placed six (later eight) identical 20-ton gadolinium-loaded liquid scintillator detectors in three underground halls. Two halls were “near” sites at baselines of 360–525 metres from the nearest reactors, used to measure the unoscillated flux. One hall was a “far” site at 1600–1900 metres baseline, where the oscillation-induced deficit would appear.

The detectors were installed at modest depth — 100 to 350 metres water-equivalent overburden, depending on location. This was much shallower than, say, KamLAND or Super-Kamiokande, but the reactor source is so intense that cosmic-ray backgrounds were tolerable with good muon vetoing.

The signature and the systematics

The detection reaction was inverse beta decay: ${\bar{\nu }}_e+p\to e^{+}+n$ with the delayed-coincidence signature of a positron (prompt signal from scintillation and annihilation gammas) followed by a neutron (delayed signal from capture on gadolinium, releasing ~8 MeV of gamma radiation). The gadolinium loading of the scintillator shortened the capture time to ~30 microseconds and produced a high-energy gamma cascade that was cleanly distinguishable from natural radioactivity backgrounds.

The measurement strategy was built around one idea: near-far comparison with identical detectors. The flux ratio between a near and far detector, after accounting for the different solid angles and reactor-to-detector distances, depends only on oscillation: $\frac{R_{\text{far}}}{R_{\text{near}}}=\frac{\text{survival probability at far baseline}}{\text{survival probability at near baseline}}$ Every systematic uncertainty that affects both detectors equally — reactor flux, detector efficiency, selection cuts, gadolinium concentration — cancels in this ratio. Only the difference in oscillation remains.

The detector design enforced this symmetry aggressively. The eight detectors were built from the same materials in the same vendor facility, filled from the same liquid-scintillator batch, calibrated with identical radioactive sources, and read out by identical electronics. The relative detector-to-detector efficiency uncertainty was held below 0.2%.

The 2012 result

Daya Bay began physics data-taking in December 2011 with six detectors (two pairs had not yet been installed in the far hall). The collaboration announced a preliminary result in March 2012 based on 55 days of data, and a full first-result paper in April.

The measurement was: ${\sin }^2(2{\theta }_{13})=0.092\pm 0.016(\text{stat})\pm 0.005(\text{syst})$ at 5.2 standard deviations from zero. RENO announced an independent, consistent measurement a month later, and Double Chooz confirmed the result with a different analysis approach. The three experiments together placed ${\theta }_{13}$ on firm observational ground.

In mixing-angle terms, ${\sin }^2(2{\theta }_{13})=0.092$ corresponds to ${\theta }_{13}\approx 8.8°$. This is much smaller than ${\theta }_{12}$ and ${\theta }_{23}$ (34° and 45°) but vastly larger than CKM-sector angles like $|V_{ub}|$ (0.23°). The non-zero value put leptonic CP-violation searches firmly on the 10-year experimental horizon.

Subsequent refinements

Daya Bay continued taking data until 2020, accumulating almost a decade of exposure. The final result, published in 2022, gave: ${\sin }^2(2{\theta }_{13})=0.0856\pm 0.0029$ — a 3.4% relative precision, the best determination of any neutrino mixing angle. The experiment also measured the reactor antineutrino spectrum with unprecedented precision, revealing an unexpected feature: a ~5-6% excess of events near 5 MeV (“the 5-MeV bump”) that does not match predicted reactor-neutrino spectra. This feature, later confirmed by RENO and other reactor experiments, is one of the standing puzzles in reactor-neutrino physics.

Legacy

Daya Bay’s measurement did three things that shaped the next decade.

Opened CP-violation searches. T2K and NOvA began their CP hunts immediately after, both using long-baseline ${\nu }_{\mu }\to {\nu }_e$ appearance with known ${\theta }_{13}$. By 2020 they had accumulated hints of non-maximal ${\delta }_{CP}$ at 2–3σ combined significance; the 5σ threshold is within reach of DUNE and Hyper-Kamiokande.

Enabled mass-ordering experiments. JUNO’s precision reactor spectrum measurement for mass-ordering determination relies on ${\theta }_{13}$ being non-zero — it is the coefficient of the rapid oscillation pattern JUNO is designed to resolve. Without Daya Bay’s ${\theta }_{13}$ measurement, JUNO would have no signal to resolve.

Killed flavor-symmetry models predicting θ₁₃ = 0. The old “tri-bimaximal” pattern — ${\sin }^2{\theta }_{12}=1/3$, ${\sin }^2{\theta }_{23}=1/2$, ${\sin }^2{\theta }_{13}=0$ — was incompatible with the data. Model building had to adapt, and a generation of subsequent flavor-symmetry work explicitly builds in mechanisms for generating the measured non-zero ${\theta }_{13}$.

Why this experiment matters as a template

Daya Bay is, in addition to its physics result, a methodological template for how to build a cross-systematics experiment. The strategy — identical detectors at multiple baselines of the same source, with aggressive detector-to-detector symmetry — is being reused in JUNO (reactor-neutrino mass ordering), the proposed PROSPECT and STEREO (sterile-neutrino searches), and even conceptually in DUNE’s near/far detector combination. The lesson is not specific to reactors or to ${\theta }_{13}$: if you can engineer a near-to-far ratio that cancels systematics, you can measure small effects without fighting every flux uncertainty independently.

Thirty years from now, textbooks will cite Daya Bay as the discovery of the third mixing angle. Experimentalists will also cite it as the moment when reactor-antineutrino physics became a precision tool for exploring the Standard Model.