Extended Data Fig. 6: TnC Efficiency & Background Rejection.

The evolution of the total neutron capture (TnC) selection is illustrated in terms of IBD selection efficiency (solid lines), total background rejection (dotted lines) and the accidental background rejection (dashed lines). The estimation of the total background rejection uses 17 days of 0-reactor data. The average singles rate per detector is ~ 10 s⁻¹. The first criterion corresponds to a time of [0.5,800]μs as a ‘loose’ coincidence with a [1.0,20.0] MeV prompt and the [1.3,10.0] MeV delayed triggers. The rates are 2291 day⁻¹ (far detector) and 2375 day⁻¹ (near detector), which imply a rejection factor of ~ 375 relative to singles. These numbers provide an absolute scale to the all other shown below. The Δr^{prompt–delay} ≤1.2 m condition yields some important reduction. However, major accidental background rejection is only obtained by the ANN with a ~ 400 rejection factor. After the ANN, the challenging correlated cosmogenic background dominates the total background rate, as expected due to the shallow overburden. The far detector is better shielded. Extra rejection uses the cosmogenic vetoes. The overall rejection factors are ~ 193 (far detector) and ~ 34 (near detector) relative to the loose coincidence.