Etude et impact du bruit de fond corrélé pour la mesure de l'angle ...

5.3 OV Tag analysis 115
The t cut value is chosen to be 10 µs (⇠ 4.5⇥⌧ SM ) as to obtain ⇢(t cut ) & 90 %
for both samples. The quantities a↵ected by the approximation performed
in Eq. 5.1 are the FN e ciency and the SM purity, since resulting from
the integral of Eq. 5.1 at short t. Using the ¯⌫ e sample to estimate the
fraction of neutrons captured at t < 10 µs, the FN e ciency and the
SM purity results underestimated by approximately 4 %. Such correction is
neglected for the SM, since its value results within the statistical uncertainty
on the SM purity. Since ¯⌫ e and FN present the same t distributions, the
fraction of ¯⌫ e events with t > t cut is assumed as FN e ciency. In this
way, the FN e ciency is known with better statistical uncertainty, since
¯⌫ e sample statistics is bigger than the FN one, and do not su↵er from the
approximation performed in Eq. 5.1. Final values of e ciency and purity
are summarised in Tab. 5.1.
tel-00821629, version 1 - 11 May 2013
Sample ✏ t ⇢ t
Stopping Muon 0.99 +0.01
0.17 0.88 ± 0.07
Fast Neutron 0.78 ± 0.01 0.97 +0.03
0.08
Table 5.1: FN an SM separation e
cut t cut = 10 µs.
5.3 OV Tag analysis
ciency (✏ t ) and purity (⇢ t ) for the chosen
The correlated background could be estimated by tagging its parent muons
crossing the OV (OVT). Since the OV veto (OVV) applied in the signal
selection, such tagged sample represent the events excluded from the signal
selection. For this reason, this method do not help to measure the
total correlated background rate, but is useful to infer its spectral shape in
[0.7, 12] MeV. The OVT is then performed by selecting those events rejected
by the OVV, i.e. time coincidences between the prompt candidate and an
OV hit in a time window of 224 ns.
In Fig. 5.4 the prompt energy spectrum in [0.7, 30] MeV is shown with
(red points) and without (blue point) OVT, a rather flat spectrum is observed.
The OVT e ciency is estimated by comparing the high energy tail
> 12 MeV with and without OVT. Assuming a flat distribution, the total
OVT e ciency is estimated to be 55 ± 6 % [45].
The FN and the SM components are then separated by the t cut between
prompt and delayed events and evaluated separately. The OVT e ciency
is estimated to be (38 ± 7) % for FN and (74 ± 12) % for SM, in agreement
with the total OVT e ciency. As expected, the OVT e ciency for SM is