This thesis presents a search for resonant production of top-antitop quark pairs in final states containing at least one electron or muon. A number of beyond the standard model (BSM) theories incorporate a specific role for the top quark, resulting in resonances that preferentially decay to t(bar over t) pairs. The data sample analysed corresponds to an integrated luminosity of 2.05 fb^-1 recorded during 2011 using the ATLAS detector at the Large Hadron Collider. The proton-proton centre-of-mass energy was 7 TeV. The search is tailored towards heavy resonances at the TeV-scale which therefore decay to top quarks with high transverse momentum. Large hadronic jets are used to reconstruct the energy carried by the hadrons and the substructure of these jets is studied in order to identify hadronically decaying top quarks. The reconstruction can therefore proceed even if the decay products of the top quark are highly collimated. This study represents the first use of jet substructure techniques in a search for t(bar over t) resonances using hadron-hadron collision data. The invariant mass of the reconstructed t(bar over t) pair is used to test compatibility of the data with the standard model prediction. No evidence for t(bar over t) resonances is found. Upper limits are derived on the production cross-section times branching ratio for narrow and wide resonant states, at the 95 % credibility level. An upper limit of 0.61 (0.65) pb is set for a narrow (wide) resonance with a mass of 1 TeV. Two specific BSM models are excluded within certain mass ranges: the narrow leptophobic Z' boson with mass 600 - 1150 GeV and the wide Kaluza-Klein gluon with a mass below 1.5 TeV. These results represent a significant improvement on those of previous searches performed at the ATLAS experiment, which did not use jet substructure techniques. In addition, the possibility of using jets which decrease in size as their transverse momentum increases is investigated using simulated data. The yield of events due to resonant t(bar over t) production increases by approximately 20 % compared to when using jets of fixed size. Furthermore, the resolution of the invariant mass of the reconstructed t(bar over t) pair is found to improve by almost one fifth. It is recommended that a calibration scheme be developed for these variable-sized jets, so that their potential to improve the sensitivity to t(bar over t) resonances can be investigated further.

This thesis is concerned with experimental investigations of properties of the top quark and processes involving this particle. In the first part of the thesis, the first determination of the electric charge of the top quark is presented. The measurement was made using top quark pair events produced in proton-antiproton collisions recorded by the D0 detector at the Fermilab Tevatron. It is based on the reconstruction of the charge of the top quarks decay products from the dominant decay to a W boson and a b-quark. The method uses a jet charge algorithm, calibrated with data, to discriminate between b- and antib-quark jets. A constrained kinematic fit is also performed to resolve the ambiguities of the pairing of the top quark decay products and to extract the top quark electric charge. The result is in good agreement with the Standard Model top quark electric charge of 2e/3 and an upper limit of 0.8 at 90\% confidence level on the fraction of exotic quarks with charge 4e/3 in the data sample is obtained. The second part of the thesis concerns the estimation of the top quark pair background to searches for new physics, such as supersymmetry, with the ATLAS  experiment at the CERN Large Hadron Collider. These searches will require a robust estimation of standard model backgrounds in order to make any claims of discovery or to exclude models of new physics. For searches with a final state signature characterized by two isolated charged leptons, multiple jets and large missing transverse energy the largest source of background is expected to be top quark pairs with leptonic decay of the two W bosons from the top quarks in the event. A data-driven method to estimate this contribution based on full kinematic reconstruction of the top quark pair events is studied using simulated proton-proton collisions. It is shown that the method is capable of estimating the top quark pair background to within 12% using data corresponding to approximately 1fb-1. The systematic uncertainty is of the order of 20% and, depending on the model, the contamination of signal events can potentially be large. / QC 20100730