Development of silicon pixel sensors for the High Luminosity upgrade of the CMS experiment at LHC and search for Higgs boson pair production in the $b\bar b \tau^+\tau^-$ final state at $\sqrt{s}$ = 13 TeV

My research activity during the PhD has focused mainly on the development of new silicon pixel sensors for the upgrade of the CMS experiment inner tracker in view of the CERN LHC High Luminosity phase (HL-LHC), an activity I started to work on during my master thesis. After a couple of years working on the characterization of the new prototypes I became interested also on the impact of the pixel detector on the physics program of the experiment. At the beginning of my second year of PhD I started to look for a physics analysis where the information of the pixel detector is largely exploited. I chose to work on the search for double Higgs boson production in final states with 2 b-jets and 2 tau leptons since the reconstruction of the b-jets and tau leptons makes large use of vertexes and tracks information. Another reason why I chose this analysis is that it will benefit from the incremented statistics foreseen at HL-LHC, possibly leading to the first experimental evidence of HH production. \vspace{3mm} \textbf{Instrumentation work} \vspace{3mm} A joint italian ATLAS-CMS INFN group has been collaborating with the Fondazione Bruno Kessler (FBK) foundry, in Trento, since 2015, to the development of a new silicon pixel sensors for the CMS experiment to be used during the high luminosity phase of the LHC collider at CERN (HL-LHC). The new pixel detector of the CMS experiment is designed to be located at $\approx 3 \; cm $ from the interaction point. At such a close distance, the irradiation fluence, after $ 2500 \; fb^{-1} $ of collected data, i.e. after seven years of operation, is expected to reach $ 2.0 \times 10^{16} \; n_{eq}/cm^2 $. The current planar design of the sensors, no matter which way they are operated, is ultimately limited by the degradation of the signal-to-noise ratio and can be reliably employed, in the best case, only up to few $ 10^{15} \; n_{eq}/cm^2 $. Therefore, a new high-radiation tolerant sensor design, capable of surviving up to ten times such a fluence, needs to be studied. Our collaboration developed thin planar and 3D pixel sensors prototypes on n-in-p wafers, employing a recent technology called Direct Wafer Bonding (DWB). Using this technology, every fabrication process takes place on one side only of the wafer, with consequent cost savings. The active thickness of the first prototypes is 100 $\mu m$ or 130 $\mu m$ for the planar sensors and 130 $\mu m$ for the 3D ones, while the pixel cells dimensions are the same of the sensors currently installed in CMS, $ 100 \times 150 \; \mu m^2 $. Several test beam campaigns have been carried out in order to characterize planar and 3D sensors coupled to the readout chip (ROC) currently used in CMS (PSI46dig), before and after irradiation. The radiation tolerance of the PSI46dig ROC limited our studies to the irradiation fluence of $ 5 \times 10^{15} \; n_{eq}/cm^2 $, which is much lower than the expected fluence at HL-LHC. Despite this limitation the results we obtained have been important to address the next step of the development program and have been published at the beginning of 2020 on a peer reviewed journal. My contribution to the chracterization of these prototypes are reported in this thesis, together with the software development I worked on. Since the second half of 2018 the first prototype of the phase 2 ROC, named RD53A, has become available. This ROC can sustain higher radiation fluences with respect to the PSI46dig, can be operated at a threshold lower than 1200 prior irradiation. The dimensions of the cell have been reduced to $50 \times 50 \; \mu m^2$ in order to cope with the higher track multiplicity expected at HL-LHC. The sensors' pixel cell area has also been reduced by a factor 6, leading to $50 \times 50 \: \mu m^2$ e $25 \times 100 \: \mu m^2$ sensors. The characterization of sensor prototypes bonded to this ROC is reported in this thesis. The goal of the future test beam campaigns is to quantify the cross talk in the $25 \times 100 \; \mu m^2$ pixel cell sensors that is caused by the bonding scheme of the sensor to the ROC imposed by the different cell area. It is important to mention that cross talk effect is present in planar sensors but is absent in 3D sensors. This would advise in favor of the use of 3D sensors in inner layers of the detector, together with the their higher radiation tolerance with respect to the planar sensors. \vspace{3mm} \textbf{Physics analysis work} \vspace{3mm} Double Higgs searches play a fundamental role in the characterization of the Higgs boson as they represent the favorite channel to measure the Higgs boson trilinear self coupling $\lambda_{HHH}$. Only three parameters shape the Higgs field potential in the Standard Model: the Higgs boson mass ($m_H$), the vacuum expectation value and the Higgs trilinear coupling ($\lambda_{HHH}$). The last one has not been measured yet. Any deviation from the theoretical predictions of the Standard Model would lead to sizable changes in both the kinematics and production rate of HH events, thus making double Higgs searches sensitive to new physics effects. The $b\bar{b}\tau^+\tau^-$ final state represents one of the most interesting channels to explore double Higgs production, because of the high branching ratio and the relatively small background contamination. At the same time, however, this final state poses some non trivial experimental challenges such as the reconstruction of the tau lepton decays that involves undetectable neutrinos, and the discrimination of signal events from background contributions. The analysis of the data collected by the CMS experiment during the LHC Run 2 (2016 + 2017 + 2018), corresponding to an integrated luminosity of 137 $fb^{-1}$ is reported in this thesis. The analysis include the study of Gluon Fusion (GGF) and Vector Boson Fusion (VBF) production channels. The internal review of the analysis started in December 2020: at this stage the statistical analysis is carried out without the inclusion of the observed data (\emph{blind analysis}), for this reason only expected results are reported in this thesis. The 95\% Confidence Level upper limit on the total (GGF + VBF) cross section is 10.7 fb (equivalent to 4.5 times the SM value). The new techniques employed in the analysis thus allowed for a major improvement of the previously published result (30 times the SM value for GGF production) based on the analysis of the 2016 dataset only. The studied of the VBF process, carried out for the first time in this analysis, resulted in an exclusion limit on the production cross section of 238.2 fb (138 times the SM values). The difference between these two results is determined by the difference in the cross sections that amounts to a factor 20. Given the limited statistics of each final state of the HH pair decay the results of all the channels will be combined in order to produce the most stringent exclusion limits for the parameters describing the production of Higgs bosons pairs.