Research Highlights Archive

KamLAND: Studying Neutrino Oscillation with Reactors

Prepared by Patrick Decowski and Lauren Hsu of the KamLAND collaboration for the DNP webpage

Since the 1950's physicist have been using nuclear reactors to study the properties of anti-neutrinos. In 1955, the first such experiment, Poltergeist, proved the existence of anti-neutrinos (as distinct from neutrinos) and measured the anti-neutrino proton capture cross section. The original Poltergeist experiment was located only a few meters away from the reactor core, the source of the anti-neutrinos. In the years since that first experiment, reactor neutrino experiments have steadily increased their baselines, with the goal to test and ultimately find neutrino disappearance. That goal was reached in 2002, when KamLAND reported the first observation of reactor anti-neutrino disappearance at an effective baseline of ~180 km 1.

Nuclear reactors produce electron anti-neutrinos (νe) in the decays of radioactive fission products in the nuclear fuel. Like the intensity of light from a light bulb or a distant star, the isotropically emitted νe flux decreases as 1/R2 for increasing distance R from the reactor. If neutrinos have mass, however, they may "oscillate" into flavors that an experiment may not be able to detect, leading to a further dimming, or "disappearance", of the electron anti-neutrinos.

Neutrino oscillation arises when the neutrino flavor eigenstates (these are the observable states) are not the same as the neutrino mass eigenstates (the states where neutrinos have definite mass). For reactor experiments, the survival probability of a νe with energy Eν after traveling a distance L from the reactor is given approximately by:

P(νe→νe) = 1 - sin2 2θ sin2(1.27 [Δm2(eV2) L(km)] / Eν(MeV)),
where Δm2 = |m12 - m22| is the difference of the mass-squares of the two mass eigenstates that are responsible for generating the oscillation (also called the mass splitting) and θ is the mixing angle between the two neutrino mass eigenstates. Notice that the absolute mass of the neutrinos does not matter in the survival probability.

The νe spectrum emitted by commercial reactors can be calculated with 2-3% uncertainty based on the fission rates of the heavy isotopes being processed in the reactor core; the fission rates are provided by the power companies operating the reactors. The accuracy of this calculation was verified in previous short baseline experiments. The average reactor νe energy is 4 MeV.The low energy of reactor anti-neutrinos makes the experiments especially sensitive to low values of Δm2. In addition, since the oscillation probability function depends explicitly on Eν, any oscillatory behavior should also manifest itself in a distortion of the neutrino energy spectrum.

The KamLAND Experiment

The Kamioka Liquid-scintillator Anti-Neutrino Detector (KamLAND) experiment is situated in the old Kamiokande cavity in a horizontal mine drift in the Japanese Alps. The site is surrounded by 53 Japanese commercial power reactors, at a flux weighted average distance of ~180 km from the reactors. This baseline makes KamLAND sensitive to the neutrino mixing associated with the large mixing angle (LMA) solution to the solar neutrino problem.

KamLAND consists of an 18 m diameter stainless steel spherical vessel with 1879 photomultiplier tubes mounted on the inner surface. Inside the sphere is a 13 m diameter nylon balloon filled with liquid scintillator. Outside of the balloon, non-scintillating, highly purified oil provides buoyancy for the balloon and acts as a shield against external radiation. Surrounding the stainless steel vessel is a water Cherenkov detector, which acts as a muon veto counter and provides shielding from radioactivity in the rock.

Electron anti-neutrinos are detected via the inverse β-decay reaction, νe + p → e+ + n, which has a 1.8 MeV νe energy threshold. The prompt scintillation light from the e+ gives an estimate of the incident anti-neutrino energy, Eν = Eprompt + <En> + 0.8 MeV, where Eprompt is the prompt event energy including the positron kinetic energy and the e+e- annihilation energy. The quantity <En> is the average neutron recoil energy, which is only a few tens of keV. The neutron captures on hydrogen ~200μs later, emitting a characteristic 2.2 MeV γ ray. This delayed coincidence signature is a very powerful tool for distinguishing anti-neutrinos from backgrounds produced by other particles.

To compensate for the loss in νe flux due to the long baseline, KamLAND has a much larger detection volume compared to earlier experiments. The KamLAND experiment uses a 1 kton detection mass, two orders of magnitude bigger than the previous largest experiment. However, the increased volume of the detector also demands more shielding from cosmic rays, which effectively means that the detector has to be placed underground.

Recent Results

Anti-neutrinos Disappear
Anti-neutrinos DisappearKamLAND started data taking in January 2002, and with only 145 days of data, reported its first results [1]. Without neutrino oscillation, the experiment expected to see 86.8±5.6 events, with 2.8 background events after all event cuts. However, only 54 events were observed. KamLAND recently confirmed this result with a 515 day data sample [2], when 365.2±23.7(syst) events were expected in the absence of oscillation, while 258 events were observed (with 17.8±7.3 background events). This establishes anti-neutrino disappearance at the 99.998% significance level.

The figure below shows, for all reactor anti-neutrino experiments to date, the survival probability of anti-neutrinos as a function of the baseline between the reactor and the detectors. The solid line and the shaded region shows the predicted survival probability from parameters determined by solar neutrino experiments, before the KamLAND results were available. There is good agreement between the measured KamLAND survival probability and the predicted value.

Anti-neutrinos Oscillate
Anti-neutrinos OscillateThe KamLAND detector not only measures the total number of anti-neutrinos, but also measures their energy. The shape of this spectrum carries additional information that can be used to investigate the neutrino oscillation. The following figure shows the measured anti-neutrino spectrum in the 515-day data sample.

The unoscillated spectrum is indicated in gray, while the data points show the measured spectrum; experimental backgrounds are also indicated in the figure. Different oscillation hypotheses are investigated by fitting them to the data. Statistical tests show that the distortion of the spectrum is inconsistent with the no-oscillation hypothesis and is also inconsistent with two alternative neutrino disappearance mechanisms, namely the decay and decoherence models. However, the spectrum is consistent with neutrino oscillation and a fit provides the values for the Δm2 and θ parameters. Since KamLAND measures Δm2 most precisely and the solar experiments exceed KamLAND's ability to measure θ, the most precise oscillation parameters are obtained by combining the results from solar experiments and KamLAND. Such a combined fit gives Δm2=7.9+0.6-0.5x10-5eV2 and tan2θ=0.40+0.10-0.07, the best solar neutrino oscillation parameter determination to date.


The KamLAND experiment has shown that electron anti-neutrinos disappear on their journey from their originating reactor to the detector. KamLAND's latest results show a distortion in the spectrum that is consistent with neutrino oscillation and strongly disfavors other disappearance mechanisms. The KamLAND experiment continues to take data and will provide the most precise determination of Δm2 in the forseeable future.


  1. K. Eguchi et al. (KamLAND Collaboration), Phys. Rev. Lett. 90, 021802 (2003).
  2. T. Araki et al. (KamLAND Collaboration), accepted by PRL, [arXiv:hep-ex/0406035].

Related Link

Neutrino Oscillation

KamLand Dectector