The NA62 experiment is focused on precision tests of the Standard Model by studies of rare decays of charged kaons since only for few kaon decays there is a precise theoretical prediction. One of these exceptions is the very rare decay K⁺→π⁺ν ν, which has a branching ratio of the order of 10^-10. Kaons rare decays such as K_L ->π₀ν ν  and K⁺→π⁺ν  ν are completely dominated by short distance dynamics. They are particularly clean in that no long-distance contributions from processes with intermediate photons are involved and hadronic matrix elements can be obtained from branching ratios of leading K decays such as K->π e v.  The branching ratios for these decays are sensitive probes of the flavour structure of the Standard Model, providing constraints on the CKM unitarity. Moreover, simultaneous measurement of the branching ratios of these two channels would provide useful information in the determination of the CKM matrix elements in a complementary and independent manner with respect to inputs from the study of B decays.

Possible contributions from physics beyond the Standard Model would easily be seen, provided a sufficient number of measured decays. Existing theories of physics beyond the Standard Model predict a significant enhancement of the decay rate and several scenarios have been proposed that would give sizable differences in both K⁺→π⁺ν ν and K_L - > π₀ ν ν branching ratios. However, so far only 7 events of  K⁺→π⁺ν ν have been reported by the E787/E949-Experiment at Brookhaven, which can only set coarse limits on possible non-SM contributions. All of them are compatible with the SM prediction but they induce a large uncertainty on the measured BR= 1.73 x 10^-10 and there is still plenty of room for possible New Physics effects.

The tiny branching ratios present an experimental challenge as these decays are extremely difficult to measure. One of the main goals of the NA62 experiment is the determination of this product with a precision better than 10%. NA62 aims at measuring a number of 100 events in about 2 years of data taking starting in 2014 with a signal acceptance O(∼ 10%) and a background of the order of 10%. To this purpose at least 10¹³ K⁺ decays are required assuming an acceptance of 10%. To obtain a contribution of the background below 10%, a rejection factor of 10¹² in rejecting the other kaon decay modes is needed and this is what drives to a large extend the design of the NA62 experiment.

Physical Motivation for RICH

In the NA62 experiment, kaons are produced by protons from the CERN SPS hitting a fixed target. Down-stream of the target, an achromatic system selects charged particles with 75GeV/c momentum which are then measured by a silicon pixel detector working at a rate of 1 GHZ (GigaTracker). In the final setup, a 400 GeV SPS primary proton beam will hit a beryllium target; producing an unseparated 75 GeV, 800 MHz beam.  Only 6% of these particles are kaons that decay in flight within a 80 m long vacuum vessel. The pion track is measured in a spectrometer, made by 4 straw chambers, two before and two after a dipole magnet, which have to operate in vacuum in order to reduce multiple scattering. Background arises from mainly two decay channels: K⁺ → π⁺ π⁰   and K⁺→ μ⁺ν , which together are responsible for about 87% of all kaon decays. In order to suppress the π⁺ π⁰ decay and be able to measure the BR(K+ -> π+ ν ν ) several sets of photon veto-counters are used to reject all events with additional photons and this results to a rejection factor of 10^-5. One of these detectors are the LAV, made by several rings of lead glass crystals around the vacuum vessel, and dedicated to detect photons which leave the detector on the outside. In addition a liquid-krypton calorimeter (LKr) that has already been used in the NA48 is doing the main rejection. Because the photon to be vetoed are two, a total rejection factor of 10^-8 can be achieved; a further 10^-3 factor is obtained by kinematics cuts. The rejection of the other channel, K⁺ -> μ⁺ ν that accounts for 63% of all kaon decays is done kinematically (10^-3) and using the absorption capability of a Muon Veto (another 10^-5 rejection factor). Finally another 10^-2 to the rejection factor is added from particle identification and the Ring Image Cherenkov counter plays an important role in adding this needed amount in order to achieve the desired rejection power of 10^-12.

The Ring-Imaging Cerenkov Counter is used to identify daughter pions from muons and electrons in an efficient and non-destructive way and it is situated between the last Straw Station (210 m from the target) and the last Large Angle Veto station.   The RICH detector provides positive identification of the Muon in the background events  and an identification of the Pion in the signal events.  The π/μ separation is crucial for achieving sufficient rejection and more than three standard deviations of π/μ separation should be achieved in the K⁺→π⁺ν ν  pion momentum range, 15 <p_π <35 GeV with a muon suppression factor which is currently better than 10^-2. In addition the RICH detector can measure pion crossing time with a resolution of 100 ps per track. The very good time resolution is exploited to match the decay information with the kaon passed inside the GTK.

This performance can be obtained by using a 17-m long, 3-m diameter volume filled with 1 atm Ne gas acting as Cerenkov radiator. The requirement for covering a specific momentum range from 15 to 35GeV leads to a reasonable compromise between the number of produced photoelectrons and the available space in the NA62 layout between the last straw chamber and the LKR calorimeter and is achieved with a gas container not longer than 18 m in the beam direction.

A series of mirrors along the downstream side of the volume will focus rings of Cerenkov light into two separated regions on the upstream side. These regions are instrumented with 2000 photomultiplier tubes (PMT’s) packed with a minimum distance of 18 mm. In a dedicated test beam for a prototype with 400 PMT’s a muon rejection better than 1% was measured with an overall pion loss of few per mil and a time resolution better than 100 ps, in the momentum range of interest for these measurements.

Picture of the mirror system

RICH Mirror layout   

To avoid the absorption of the reflected light on the beam pipe, the Cherenkov photons emitted by pions and muons are reflected by two semi-spherical mirrors: one with the centre of curvature to the left and one to the right of the beam pipe. Since the area to be covered by each semi-spherical mirror is very large (~ 3 m2 ), a mosaic of smaller segments is used. A total of 18 hexagonal and 2 semi-hexagonal mirrors, each inscribed inside a 70 cm diameter circle, have been provided by  the MARCON company.

The mirrors are made of thick glass (2.5 cm) with a focal length of 17 m and an exceptional optical quality (D0 < 4 mm, where D0 is the diameter of the circle containing the 95% of the imaged light from a point source placed in the centre of curvature). The glass is coated with aluminium and covered with a thin dielectric film (MgF_2) to protect the reflecting surface. The average reflectivity is > 90% in the wavelength range between 195 and 650 nm. Each mirror must be supported and adjustable for alignment. A dowel with a spherical head is inserted in the hole drilled almost in the centre on the not-reflecting surface and used to sustain the mirror. Two actuating ribbons at ± 45° with respect to the vertical keep the mirror in equilibrium and give the alignment. Piezo motors with a 35 mm range and 70 nm resolution, located out of the acceptance, are used to pull the ribbons counter-balanced by the mirror weight.  A third stabilizing ribbon avoids the rotation of the mirror around the dowel axis. An aluminium honey-comb structure, 50 mm thick, has been chosen as mirror support framework to minimize the material in front of the electromagnetic calorimeter.

The Photomultipliers 

Metal packaged Photomultipliers of the HAMAMATSU R7400 series of the U-03 (UV glass window) type were chosen for their compactness and fastness after the comparison tests carried-on during the 2009 test run. The R7400 PM has a polyoxymethylene insulation cover of roughly cylindrical shape, 15.9 ± 0.4 mm wide and 11.5 ± 0.4 mm long.

The eight PM dynode voltages are provided through a custom made HV divider (28 MΩ total resistance), which has a cylindrical shape and three cables: one for signal output and two cables for high voltage (negative) supply and for grounding (see Figure 1).

Figure 1: The HAMAMATSU R7400 U-03 photomultiplier connected to the custom made divider.

The photocathode (with 8 mm minimum active diameter) is bialkali made and has a typical radiant sensitivity of 62 mA/W at the 420 nm peak wavelength, corresponding to a 20% quantum efficiency; the PM wavelength sensitivity is between 185 and 650 nm. The typical PM gain is 1.5 x 10⁶ at 900 V (working voltage, maximal safe HV is 1000 V).

The R7400 typical rise time is 0.78 ns, the transit time is 5.4 ns and the transit time jitter is 0.28 ns (FWHM). This results into a RICH resolution better than 100 ps for pions with momentum between 15 and 35 GeV/c as directly measured during the 2009 test run[1] (see Figure 2).

Figure 2:  Time resolution as a function of pion momentum. The time resolution is better at higher momenta due to the increasing number of firing PMs.

The PMs will be placed in the upstream end-cap of the RICH, in the region where the pion Cherenkov rings are expected. Each PM will be placed into a cylindrical cavity, 16.5 mm wide.

In the inner part of the housing flanges, for each PM a truncated cone is drilled, 18 mm wide at the beginning, 7.5 mm wide at the end and 22 mm high (see Figure 3), in order to convey the Cherenkov light to the active area of the PM (Winston cone approach [2]).

The cone is covered with an aluminized Mylar foil, 50 mm thick, glued to the aluminium surface, in order to improve the reflectivity. A 1 mm thick, 12.7 mm wide quartz (fused silica) window is glued between the cone and the cylinder in order to separate the PM from the neon (RICH radiator).

Figure3: section of a single optical detection module (measurements are in mm); the lower part shows the Winston’s cone and the place for the quartz window, while the upper one shows the cavity housing the PM and the voltage divider.

The PMs will be powered by four CAEN SY2527 crates equipped with A1535S boards, each one providing 24 HV channels.  In order to reduce the number of modules, each HV channel will power 4 PMs.

The data acquisition system will be based on the following components:

• custom made 32 channels boards, each one housing 4 NINO[3] ASIC discriminator chips; before the discriminators the signals will be amplified and converted to a differential output in the same board.
• 128 channels TDC daughter Boards[4] (TDCB), each one housing 4 CERN HPTDC[5] (High Performance TDC). The TDCB will receive the LVDS signals from the NINO discriminators.
• TEL62[6] mother boards (developed on the basis of the LHCb TELL1[7] board), each one housing 4 TDCB, that will process the digital information and produce both the data stream to be sent to the online PC farm and the trigger primitives that will be evaluated by the Level 0 trigger processor (L0TP).

The RICH neon container (vessel) is presently under construction and  is expected to be delivered to CERN in September 2013; it is a major piece of mechanics, given its dimensions  (diameter increasing from 3 to 4 m, overall length of 17 m, divided  into 4 sections) and the requirement to be able to be fully evacuated  before injecting fresh neon. The vessel will be installed and vacuum  tested in the NA62 cavern in the fall 2013.

In the first half of 2014 the mirror system will be installed inside the RICH vessel and later the photomultipliers will be placed in their  final position. In the summer 2014 the RICH detector will be 
completed, in time for the first run of the full NA62 experiment  foreseen for the fall 2014.

The RICH project is a collaboration of INFN sezione di Firenze, INFN Sezione di Perugia, University of Perugia, University of Firenze and CERN.

People involved are:

G.Anzivino (INFN-PG and Univ-PG),

F.Bucci (INFN-FI),

V.Carassiti (INFN-Ferrara),

A.Cassese (INFN-FI and Univ-FI),

P.Cenci (INFN-PG),

R.Ciaranfi (INFN-FI),

V.Duk (INFN-PG),

E.Iacopini (INFN-FI and Univ-FI),

S.Lami (INFN-FI),

M.Lenti (Project leader, INFN-FI),

M.Pepe (INFN-PG),

M.Piccini (INFN-PG),

P.Wertelaers (CERN)

The author would like to thank Massimo Lenti, Mauro Piccini and Francesca Bucci for kindly providing information about the project and for carefully reading the original draft of this article. ​

Further Readings:

[1] B. Angelucci et al., Nucl. Instrum. Meth. A 621 (2010) 205.

[2] R. Winston, J. Opt. Soc. Am. 60 (1970) 245.

[3] F. Anghinolfi, P. Jarron, A.N. Martemiyanov, E. Usenko, H. Wenninger, M.C.S. Williams, A.

       Zichichi, Nucl. Instr. and Meth. A 533 (2004) 183.

[4] Collazuol, G et al., Proceedings of Nuclear Science Symposium Conference Record

       (NSS/MIC), 2009 IEEE (2009) 1138.

[5] M. Mota, J. Christiansen, A flexible multi-channel high-resolution time-to-digital converter

       ASIC, Proceedings of 2000 Nuclear Science Symposium and Medical Imaging Conference

       Lyon, vol. 1520, 2000.

[6] B Angelucci, E Pedreschi, M Sozzi, F Spinella, Proceedings of Topical Workshop on

       Electronics for Particle Physics 2011 (TWEPP-11) JINST 7 C02046 (2012).

[7] G. Haefeli, A. Bay, A. Gong, H. Gong, M. Muecke, N. Neufeld, O. Schneider,  Nucl. Instr. and

       Meth. A 560 (2006) 494.