The RICH detector of the NA62 experiment

The NA62 experiment at CERN is aimed at measuring the ultra-rare decay K+→π+νν with 10% accuracy. One of the detectors that is crucial for the rejection of background events is the RICH detector: a gas based detector aimed at π/μ separation in the 15–35 GeV/c momentum range with an inefficiency of less than 1%. The RICH must also provide a very precise time measurement (with the time resolution ∼100 ps) to correctly associate the π+ with the parent K+ particle measured by an upstream detector. This paper contains the detailed description of the RICH detector, its readout, and the results of the commissioning run at CERN in 2014.


Abstract.
The RICH detector of the NA62 experiment is one of the key detectors to achieve the muon rejection needed in the search for the K + → π + νν decay, performed by NA62. Since BR(K + → µ + ν) is higher than BR(K + → π + νν) by more than 9 orders of magnitude, it represents one of the most relevant background contributions. Its rejection is performed with kinematic reconstruction of the event and identification of the charged particles in the final state (π + against µ + ). The pion identification efficiency using the RICH detector is measured to be 83% in the momentum range between 15 and 35 GeV/c, with a misidentification probability for muons of 0.2%, while the track crossing time is measured with a resolution of 70 ps. The RICH detector is also exploited to provide trigger for charged particles with an efficiency greater than 99%.  1. NA62 and the K + → π + νν decay The ultra-rare K + → π + νν decay is a good environment to test the Standard Model in the Flavor Physics sector, its branching ratio is precisely computed in the Standard Model [1]: BR SM (K + → π + νν) = (0.84 ± 0.10) · 10 −10 . In addition, it is sensitive to several New Physics models.
The branching ratio measurement has already been performed by the BNL E787 and E949 experiments using kaon decays at rest [2] BR BN L (K + → π + νν) = (1.73 +1.15 −1.05 ) · 10 −10 . The goal of NA62 is to perform the measurement of the BR with a 10% precision, comparable with the Standard Model accuracy.
NA62 collected data in 2016, 2017 and 2018, and needs to run after CERN LS2 period to complete the measurement. The analysis of 2016 data-set led to the observation of 1 signal candidate [3], while in 2017 data-set 2 signal candidates have been detected. The preliminary combined (2016 and 2017 data) results from NA62 are [4]: (1) The full description of the NA62 beam and detector (including RICH) is provided in [5].

The RICH detector
A schematic view of the RICH detector is shown in fig.1.  3 ribbon prevents rotations. A system of piezo-motors, out of acceptance and remotely controlled, allows the alignment of each mirror, with a precision of 30 µrad [6].
The Cherenkov photons are collected by two disks of 976 PMs (Hamamatsu R7400U-03) each. The PMs have 16 mm wide face, 8 mm active region, and are packed in hexagonal structure with 18 mm cell size. They are sensitive between 185 ÷ 690 nm with a maximum quantum efficiency of 20% at λ peak ∼ 420 nm. Winston cones are exploited to increase the geometrical coverage.

RICH basic performance
The RICH space and time resolution, together with other basic performance, have been measured directly on data taken during the commissioning and the first physics run [5], [6].
In order to perform unbiased measurements, a clean sample of positrons have been exploited. Since the positron mass is negligible with respect to the momentum for p > 10 GeV/c (β = 1), the positrons are always above the Cherenkov threshold, and the Cherenkov ring radius and the number of hits do not depend on the particle momentum. Moreover, only events with the expected Cherenkov ring fully contained within the RICH geometrical acceptance have been taken into account. The positron sample has been collected selecting K + → π 0 e + ν events, applying kinematic and calorimetric requirements.
The number of hits per positron ring, obtained fitting the distribution with the Poisson function, is: N hits e 13.8. The positron ring radius, obtained fitting the distribution with the Gauss function, is: R e 189.6 mm, σ Re 1.47 mm. The measurement of the single hit space resolution is performed defining the quantity P ull = (R − R exp ) (N hits − 3), in order to weight the contribution to the resolution coming from each hit. R is the reconstructed ring radius, R exp is the expected ring radius assuming the positron mass, N hits is the number of observed hits, 3 is the number of free parameters used in the χ 2 -based hits fit that reconstructs the ring (radius, X and Y center coordinates). Then σ Hit = σ P ull 4.7 mm (fig.2, left).
The ring time resolution is calculated by randomly dividing the hits of a single ring into two groups, defining the time of each group as the average hit time, and fitting T 1 − T 2 ( fig.2, right). Then, the ring rime resolution is: σ T = 0.5 · σ T 1 −T 2 70 ps.
4. RICH usage in K + → π + νν selection In the K + → π + νν selection the RICH plays a relevant role to reject background coming from kaon decays with a muon in the final state, typically the K + → µ + ν µ decay, that must be suppressed by more than 10 orders of magnitude. In this section the RICH usage in the analysis of 2017 data is described.
In the left plot of fig.3 it is shown how the RICH detector works: the ring radius measured by the RICH, versus the momentum measured by the spectrometer, gives separated distributions for different particles, since the charged particle mass can be expressed as: where p is the charged particle momentum, R is the ring radius, n is the refractive index and F the focal length. The rings are reconstructed with two different analysis methods in order to optimize the signal acceptance and the background rejection ( fig.3, right). In particular a stand-alone ring reconstruction is performed, and using the spectrometer momentum in eq. 3, the charged particle mass is computed ( fig.4, left). Independently, the expected ring center is obtained by extrapolating the track to the PM plane (reflecting the track off the mirrors as if it were a photon); this information is used to compute different ring radii for different mass hypotheses, reverting eq. 3 and a likelihood discriminant is computed for each hypothesis ( fig.4, right).
The RICH also contributes to the level-0 trigger chain [7], with an efficiency higher than 99%.  Figure 4. Left: charged particle mass reconstructed by the RICH detector, for muons (blue) and pions (red) in the momentum range 15 − 35 GeV/c; the optimized cut applied in the πνν selection (125 MeV/c 2 ) is shown in black. Right: highest likelihood discriminant of the notpion hypotheses, reconstructed by the RICH detector, for muons (blue) and pions (red) in the momentum range 15 − 35 GeV/c; the optimized cut applied in the πνν selection (0.12) is shown in black.