Evidence for $\pi K$ -atoms with DIRAC-II

DIRAC-II is a fixed-target experiment at the CERN Proton Synchroton (PS) which has been designed to search for piK atoms, a bound state of a pi±K± pair, and measure their lifetime. These atoms are observed through an excess of low energetic piK pairs over the background, detected in the two spectrometer arms. This excess comes from the ionization of piK atoms in the target and can be related to their mean life. The piK S-wave scattering length combination |a1/2 - a3/2| (for isospin 1/2 and 3/2) can be related to the latter. The aim of the upgraded DIRAC-II experiment is a measurement of the scattering length combination |a1/2 - a3/2| with a precision of 5%. piK atoms have not been observed so far. The original DIRAC experiment was designed to measure the scattering lengths of pipi atoms. So far, close to 15 000 atoms have been detected, leading to a precision on |a0 - a2| which is better than 10%. In chiral perturbation theories (ChPT) the pipi scattering lengths have been calculated with 2% precision and are in good agreement with the DIRAC-II measurements. Predictions for the piK ones, which involve the s-quarks are obtained. DIRAC-II data will be very important to check ChPT extended to the s-quark. Previous measurements of a1/2 and a3/2 suffer from very large errors as they are extrapolated from high energy piK scattering. My first contribution was the development and construction of a Cherenkov threshold count er for the separation of kaons from pions, which is crucial for the detection of piK atoms. The momentum range (4-8 GeV/c) requires aerogel with refractive index n = 1.008 and n = 1.015 as Cerenkov radiator. Because of the strong light absorption in aerogel, the low indices and the very large size of the active area needed in the DIRAC-II experiment, new designs had to be invented and investigated. In this work new concepts of aerogel counters are presented e.g. introducing wavelength shifters in order to avoid the strong light absorption in aerogel. The second contribution was the analysis of the first run in 2007 and the search for piK atoms. A new tracking had to be developed to optimize the reconstruction of tracks using only detectors downstream of the spectrometer magnet, since no upstream detectors were operational at that time. This new method was carefully tested on the already published pipi 2001 data sample. From the 1137 ± 257 observed piK Coulomb pairs in the signal region where atoms are expected, one predicts that piK atoms must have been produced. Despite the low statistic collected during the first data run in 2007, we have already evidence for 173 ± 54 piK atoms with a significance of 3.2 sigma, which is in good agreement with expectation from the number of observed Coulomb pairs. A lower limit of 1.5 fs could be given for their mean life at 84.1% confidence level.