Conclusion

Part V

Chapter 12

Conclusion

The study of neutrino oscillations provides a unique window to physics beyond SM. There have been a lot of progress in the neutrino oscillation physics in the recent years. The mixing angle θ₁₃ is finally measured in 2012. To date, we know the value of all of the three mixing angles (θ12, θ23, θ13) and the mass splittings (∆m²₁₂,|∆m²₃₂|). However, the value of CP phase δCP is still not known yet.

In this thesis, we reported the measurement of ν_µ → ν_e oscillation. This oscillation mode is particularly important because it is sensitive to the value of δCP and θ13. There are two fundamental questions related to the measurement of neutrino oscillation.

• Is there a CP violation in the lepton sector?

• What is the physics behind the mixing pattern?

Precise measurement ofνµ→νe is essential to address these questions.

In the T2K neutrino oscillation analysis, the dominant systematic error is originated from the uncertainty in the measurement of neutrino-nucleus interactions. When we measure the neutrino beam at ND280 and SK, we select the CCQE interaction (ν+N →l+N) as a signal, while the main background for CCQE is CC1π interaction (ν+N →l+N+π). Distinguishing these two interaction modes is very important in the measurement of neutrino oscillation.

The FGD is the key detector in the ND280 detector complex for distinguishing the neutrino interaction modes. The CCQE and CC1π⁺interaction modes can be distinguished by detecting the pion track in the final state. Because the FGD is made of fine-grained scintillator bars and acts as active target, it is capable of detecting the short-ranged pion tracks around the interaction vertex. The light from scintillator bars are read out by the MPPCs, which are suitable to be used in ND280 thanks to its compactness and insensitivity to magnetic fields. We developed a new track reconstruction algorithm based on the Radon transform method. Using this algorithm, the reconstruction efficiency for multiple tracks is increased by ∼15%.

In the CC1π⁺ interaction, pions are often absorbed inside a nuclei before being observed in the detector. Also,π⁺may convert toπ⁰via the charge exchange interaction (π⁺+N →π⁰+N).

When the pion absorption (ABS) or charge exchange (CX) happens, there are no π⁺ in the final state, so the CC1π event is misidentified as the CCQE event. About half of the π⁺ in the final state of CC1π interactions disappear before escaping the nuclei due to ABS and CX. Although these interactions are important, we apply a large uncertainty in the pion-nucleus interaction model, because the uncertainties in the past pion-nucleus cross section measurements used for tuning the model parameters is large. The uncertainties of π−C cross section in the past experiments is typically ∼25% for ABS and ∼50% for CX. The uncertainty in ABS and CX results in the uncertainty of the number of νe signal candidate events at SK.

In order to reduce the uncertainties in those pion interactions, we performed a pion-nucleus cross section measurement at the pion secondary beam line at TRIUMF. A scintillating fiber

Chapter 12. Conclusion

tracking detector was newly developed for this measurement. As it is super-fine-grained and fully active, we were able to select the ABS and CX interaction with high efficiency (∼80%) and high purity (∼75%). We measured the sum of ABS and CX cross section with total uncertainty of ∼6.5%. Compared to the uncertainties of the past experiments, the uncertainty is improved by a factor of 2.

By using this results together with the π−C cross section data sets from past experiments, we tuned the parameters in the pion-nucleus interaction model used in T2K. By improving the tuning method and using our own new data sets, we reduced the uncertainty of the model parameters to∼1/4. In the current T2K neutrino oscillation analysis, the improved model is not used yet. Without this improvement, the uncertainty of the model results in 2.3% error for the number of ν_e events at SK (assuming sin²2θ₁₃ = 0.1). This is one of the dominant systematic error sources in 2013ν_e appearance analysis. This error will become negligible once we use the improved model.

Compared to the T2Kν_eappearance analysis in 2012, the measurements at both ND280 and SK were improved. In ND280, we changed the event categorization that we use in theν_µ beam measurement. There were only two categories (CCQE-like, non-CCQE-like) in 2012, while in this analysis we divided the sample to three categories (CC0π, CC1π⁺and CC other), using the reconstructed tracks at TPC and FGD. With this new categorization, the systematic errors for the single pion production cross section was reduced, and the uncertainty of the predicted number ofν_e events related to the ND280 measurement was reduced to 2.9% (assuming sin²2θ₁₃= 0.1), while it was 5.0% in 2012.

In the SK νe event selection, the cut to reject the background from NC1π⁰ interaction (ν_µ+N →ν_µ+N+π⁰) was improved by using a new event reconstruction algorithm. The new algorithm defines the particle type, vertex position and momentum at the same time by using a maximum likelihood method, while in the old algorithm those parameters were defined one by one. Using the new algorithm, the NC1π⁰ background was reduced to less than half.

Using the improved measurement from ND280 and SK, we fitted the SK ν_e candidate events to extract the oscillation parameters. The data sets we used for this analysis was

∼2.2 times larger than the data sets used in 2012 analysis. The result of the fit assuming sin²θ₂₃= 0.5,∆m²₃₂= 2.4×10⁻³ eV²,δ_CP= 0 and normal (inverted) hierarchy is:

Best fit: sin²2θ13 = 0.140 (0.170)

68% C.L.: 0.108 <sin²2θ₁₃<0.178 (0.133 <sin²2θ₁₃<0.214) 90% C.L.: 0.090 <sin²2θ13<0.205 (0.111 <sin²2θ13<0.246)

The significance to exclude θ₁₃= 0 was 7.3σ for both hierarchy cases. We also performed a fit with the uncertainty of sin²θ₂₃ and ∆m²₃₂ taken into account. The constraints on sin²θ₂₃ and

∆m²₂₃ are applied by using the results from T2K Run 1-3νµ disappearance measurement.

Best fit: sin²2θ₁₃ = 0.136 (0.166)

68% C.L.: 0.103 <sin²2θ13<0.180 (0.124 <sin²2θ13<0.217) 90% C.L.: 0.084 <sin²2θ13<0.214 (0.102 <sin²2θ13<0.256)

Finally, by adding the constraint on sin²2θ13from the average of reactor measurement (sin²2θ13= 0.098±0.013), we obtained 90% excluded regions forδCP.

Normal hierarchy: 0.604 ∼ 2.509

Inverted hierarchy: −3.142∼ −3.043, −0.132 ∼3.142

Our data prefers δ_CP = −π/2. The constraint on δ_CP is still weak to claim an evidence of non-zero CP violation, but this is an important milestone in the neutrino oscillation physics.

T2K or the next generation experiments such as T2HK [141] may reveal the value ofδ_CP in the near future.