Measurement of the Higgs-Boson Production in Association with a Top-Quark Pair in the Fully-Hadronic Final State with the CMS Experiment

This thesis presents the results of the measurements of t¯tH(b¯b) with data collected by the CMS experiment between 2016 and 2018 in proton-proton collisions at 13TeV at the LHC. The main focus of the work is on the fully-hadronic final state, in which the top quarks decay exclusively to light and bottom quarks. The ATLAS collaboration published a measurement of this final state using data collected at 8TeV [9]. This channel was pioneered at CMS using 36.3 fb−1 of data collected in 2016 [10] at 13TeV. A refined version of this analysis was published in combination with other t¯tH(b¯b) final states using an additional 41.5 fb−1 of data collected in 2017 [11]. The fully-hadronic channel exhibits a total of eight jets, four coming from the hadronization of light quarks from the W-boson decays and four from the bottom quarks. Ideally, signal events would present themself in the CMS detector with eight jets, of which four could be identified as b-jets. Due to experimental reasons, events with at least seven jets are used to measure this t¯tH(b¯b) process. Since no leptons are present in the final state, the signal is hidden underneath an extensive and dominating background from QCD multijet events, the most common process at hadron colliders. Additionally, producing top-quark pairs with additional jets (t¯t + jets) originating from gluon splitting contributes a sizable background. Especially t¯t with additional bottom-quark jets (t¯t +b¯b) is a background process that cannot be easily removed due to its similarity to the signal. Separating the signal from the backgrounds is among the most crucial challenges in analyzing the fully hadronic t¯tH(b¯b) channel. Furthermore, the modeling of the QCD multijet background presents itself as a limiting factor in the sensitivity of the measurement. This thesis analyzes the fully-hadronic decay channel of t¯tH(b¯b). It directly builds on the foundations of the analyses presented in references [10, 11] by improving and refining the methods used: New dedicated triggers, targeting all-jets final states, were developed for the CMS detector's 2017 and 2018 data-taking periods. The estimation method for the QCD multijet background was revised and improved. A novel neural-network-based classifier was developed to separate the t¯tH(b¯b) signal from the QCD multijet and, to a lesser extent, the t¯t + jets background. Improved estimation of the QCD multijet background allowed us to avoid the previous statistical limitation and use collected data events in training the neural network classifier. Additionally, the analysis implemented the state-of-the-art modeling of t¯t + b¯b and an improved algorithm for b-jet identification.