Module Prototype Qualification For The Cms Pixel Detector Upgrade

Module Prototype Qualification for the CMS Pixel Detector Upgrade

Development of a New Tracker for the CMS Upgrade Phase 2 and Study of the HL-LHC Physics Reach

The standard model of particle physics provides a coherent description of highenergy physics processes and has been hugely successful in providing experimental predictions. Among its long list of achievements, the most significant is arguably that of the discovery of the Higgs boson half a century after being theorised, providing the last cornerstone needed for the standard model to become fully consistent. Despite huge successes, the standard model still suffers from major shortcomings. On the path leading towards a better understanding of particle physics, an in-depth study of the Higgs boson is key. This relentless work of characterising the properties of the Higgs boson is currently being undertaken at the Large Hadron Collider, where high-energy proton collisions are being recorded by dedicated detectors, providing a continuous improvement to the understanding of the standard model. Amid tremendous achievements, some processes, remain too weak to be detected with the current installations. One such measurement is the combined production of two Higgs bosons allowing for a direct handle on the Higgs self-coupling parameter of the standard model. To maximise the physics reach of the collider, it will be subjected to a major upgrade, allowing for a strong increase in luminosity. Such a dramatic change will bring major challenges to the experiments recording these collisions and upgrades are required if they are to maintain their outstanding performance. This thesis explores the upgrade of the CMS silicon strip detector, centred around the in-beam characterisation of detector module prototypes and discusses the physics reach of the upgraded machine, with an emphasis on Higgs boson pair production in the bbWW(l) final state.

Beam Test Calorimeter Prototypes for the CMS Calorimeter Endcap Upgrade

​In order to cope with the increased radiation level and the challenging pile-up conditions at High Luminosity-LHC, the CMS collaboration will replace its current calorimeter endcaps with the High Granularity Calorimeter (HGCAL) in the mid 2020s. This dissertation addresses two important topics related to the preparation of the HGCAL upgrade: experimental validation of its silicon- based design and fast simulation of its data. Beam tests at the DESY (Hamburg) and the CERN SPS beam test facilities in 2018 have been the basis for the design validation. The associated experimental infrastructure, the algorithms deployed in the reconstruction of the recorded data, as well as the respective analyses are reported in this thesis: First, core components of the silicon-based prototype modules are characterised and it is demonstrated that the assembled modules are functional. In particular, their efficiency to detect minimum ionising particles (MIPs) traversing the silicon sensors is found to be more than 98% for most of the modules. No indication of charge sharing between the silicon pads is observed. Subsequently, the energy response is calibrated in situ using the beam test data. Equalisation of the different responses among the readout channels is achieved with MIPs hereby deploying the HGCAL prototype as a MIP-tracking device. The relative variation of the inferred calibration constants amounts to 3% for channels on the same readout chip. The calibration of the time-of-arrival information is performed with an external time reference detector. With it, timing resolutions of single cells including the full prototype readout chain around 60ps in the asymptotic high energy limit are obtained. The calorimetric performance of the HGCAL prototype is validated with particle showers induced by incident positrons and charged pions. For electromagnetic showers, the constant term in the relative energy resolution is measured to be (0.52± 0.08) %, whereas the stochastic term amounts to (22.2 ± 0.3)% √GeV. This result is in good agreement with the calorimeter simulation with GEANT4. The prototype’s positioning resolution of the shower axis, after subtracting the contribution from the delay wire chambers in the beam line used as reference, is found to be below 0.4 mm at 300 GeV. At the same energy, the angular resolution in the reconstruction of the electromagnetic shower axis in this prototype is measured to be less than 5mrad. The analysis of the hadronic showers in this thesis makes use state-of-the- art machine-learning methods that exploit the calorimeter’s granularity. It is indicated that the energy resolution may be improved using software compensation and also that the separation of electromagnetic and charged pion-induced showers in the calorimeter may benefit from such methods. The measurements of the hadronic showers are adequately reproduced by GEANT4 simulation. Altogether, the obtained results from the analysis of the beam test data in this thesis are in agreement with the full functionality of the silicon-based HGCAL design. The final part of this thesis provides a proof of principle that generative modelling based on deep neural networks in conjunction with the Wasserstein distance is a suitable approach for the fast simulation of HGCAL data: Instead of sequential simulation, a deep neural network-based generative model generates all calorimeter energy depositions simultaneously. This genera t or network is optimised throu gh an adversarial training process using a critic network guided by the Wasserstein distance. The developed framework in this thesis is applied to both GEANT4- simulated electromagnetic showers and to positron data from the beam tests. Ultimately, this fast simulation approach is up to four orders of magnitude faster than sequential simulation with GEANT4. It is able to produce realistic calorimeter energy depositions from electromagnetic showers, incorporating their fluctuations and correlations when converted into typical calorimeter observables.