hide
Free keywords:
ILC
Abstract:
Calorimeters with a high granularity in transverse and longitudinal direction are essential for the design of particle flow focussed detector systems at future e+e−-collider experiments. The CALICE scintillator-SiPM analog hadronic calorimeter (AHCAL) is such a highly granular calorimeter. This thesis studies the high precision timing capabilities on channel level of a new generation of AHCAL prototypes and their potential for the energy reconstruction. A time calibration scheme is developed for the AHCAL using test beam data. Formuons, a time resolution of 6.3 ns is achieved with a partially equipped prototype. For electromagnetic and hadronic showers the time resolution broadens due to an electronics effect ocurring at higher occupancies in the detector. Late energy depositions, which are consistent with slow neutron events in their radial, longitudinal and hit energy distributions, are observed for hadronic showers. When comparing the amount of those late energy depositions to simulated data, significant deviations are observed. It is further investigated towhat extent the hit time measurements can be used to improve the energy resolution of the AHCAL. Strong correlations between late energy depositions in a hadronic shower and deviations of the reconstructed energy from the true energy are visible in simulated data of an AHCAL prototype. It is shown that these correlations can be exploited to enhance the energy resolution by about 15% compared to the standard energy reconstruction. However, this improvement does not outperformtraditional software compensation (SC) methods. It is further shown that global observables describing the hit energy spectrum of a hadronic shower can be used to correct the energy reconstruction in a similar way as observables constructed from the hit time measurements. Finally, a machine learning setup is developed on the basis of an artificial neural network for the energy reconstruction in the AHCAL. It is shown that the longitudinal shower profiles of hadronic showers can be used to enhance the energy resolution. Together with other global features the energy resolution can be improved by up to 40% in simulated data. This amounts to a significant enhancement over the traditional SC approach. Training the network on simulated data and subsequently applying it to test beam data leads to improvements in the energy resolution of about 30 %. The studies presented in this thesis add to our understanding of the time development of hadronic showers in calorimeters and its simulation. Furthermore, it is shown that multivariate, machine learning based methods for the energy reconstruction hold the potential for significant improvements in the energy resolution of highly granular calorimeters.
