Electron and photon energy reconstruction in the electromagnetic calorimeter of ATLAS

The Atlas LAr electromagnetic calorimeter is designed to provide a precise measurement of electrons and photons energies, in order to meet the requirements coming from the LHC physics program. This request of precision makes important to understand the behavior of the detector in all its aspect. Of fundamental importance to achieve the best possible performances is the calibration of the EM calorimeter, and this is the topic of this thesis. With detailed Monte Carlo simulations of single electrons and photons in the Atlas detector, we find a method to calibrate the electromagnetic calorimeter, based only on the informations that come from it. All the informations needed to develop a calibration method come from the simulations made with the technique of the Calibration Hits, that allows to know the en- ergy deposited in all the materials inside the detector volume, and not only in the active layer of each subdetector as possible in the standard simulations. This technique required a big effort for the development of all the algorithms, because at the time of the firsts rounds of simulations the standard recon- struction code could not work on the calibration hits informations. The simulations can be grouped into four rounds, with different conditions to disentangle different effects and allow to study in detail the showering process inside the calorimeter. First step was to study a possible calibration scheme, and we start in the simplest possible condition: electrons hitting the center of a middle compart- ment cell, without the magnetic field in the inner detector region. Detailed study on this simulations allows us to find a possible calibration procedure. In this simple condition the performances of the electromagnetic calorimeter barrel have been tested, giving good results in term of energy resolution and linearity, with a sampling term varying in $\eta$ from the 10% to the 15% and the linearity within 0.5%. Second step was to introduce the effect of the impact point of the particles inside the cell on the reconstructed energy. This was done simulating electrons with the impact point on the calorimeter spread uniformly over the full middle compartment cell, without the magnetic field in the inner detector. Studies on the reconstruction of the barycenter of the shower and its comparison with the simulated $\eta$ and $\phi$ positions have been done. Two modulations on the reconstructed energy, due to the $\eta$ and $\phi$ positions of the impact point inside the cell, are studied and two correction were included in the method. The performances of the method in term of energy resolution and linearity are slightly affected by the impact point modulations: the resolution goes from 10% to 17% in the barrel region, with a linearity better than 0.5%. The third step in our analysis was to introduce the magnetic field in the inner detector region. This is the normal condition in the Atlas experiment and need to be studied in great detail. The effect of the magnetic field on the shower development is to deflect the particles of the shower in the $\phi$ direction, and this effect can be very large for low energy electrons, that can be deflected up to 4 middle cell width. The largest effect of the magnetic field is on the electrons that emit hard bremsstrahlung photons: in this case the presence of two different clusters in the calorimeter is evident and introduce large errors in the evaluation of the energy deposited outside the cluster. This problem is solved with a new method for the reconstruction of the energy deposited outside the cluster. The method performances are a little worse: resolution from 10% to 17% and linearity within 0.5%, excluding the 5 GeV electrons. The first three steps of analysis are done on private data sets that have been simulated with big effort on the Milan computing facility. With the aim of including the proposed calibration procedure in the Athena framework, we focused on the analysis of the data available from the central production, the CSC (Computing System Commissioning) data set, that become available in 2007 and contains all the informations from the calibration hits that are required for the proposed calibration method. This CSC data sets are sim- ulated with the newest detector geometry and are digitized to simulate in the best way the real data that will be given by the detector in the actual operative condition of Atlas. Different energies have been simulated, both for electrons and photons, covering uniformly the full $\eta$ range of the calorimeter, including also the two endcap regions not simulated before for the limited computing resources available. The method proposed shows good results both for electrons and photons, and over all the $\eta$ range of the calorimeter. Finally the proposed calibration method is implemented in the Athena framework, and can be used as an alternative to the standard calibration method based on the longitudinal weights. The performances of the pre- liminary implementation of the proposed calibration method are studied, in comparison with the performances of the other calibration method. The method proved also to be not very sensitive to an underestimation of the material in front of the calorimeter: a 10% of variation of the upstream material, change the resolution for 100 GeV electrons of 2% and the linearity of 0.5%. The proposed method gives, in its first implementation into Athena, results comparable with the ones provided from the other calibration method based on longitudinal weights. The parameter extraction, and consequently the method performances, can be improved, provided that larger statistic will be available. The CERN community of liquid argon detector have largely approved this new calibration method, that is recognized as a valid alternative to the standard calibration method.