The heart of ATLAS: Commissioning and performance of the ATLAS silicon tracker

The Large Hadron Collider (LHC) has been built under the french-swiss border near Geneva, Switzerland. Two opposing beams of protons will collide with a centre of mass energy of 14 TeV, an energy seven million times that of the first accelerator. The LHC takes particle physics research to a new frontier. On September 10th 2008, the first single pilot beam of $2 x 10^9$ protons was circulated successfully through the entire LHC, with an injection energy of 0.45 TeV. The first collisions are expected in Summer 2009. One of the experiments designed to search for new particle phenomena is the ATLAS experiment. This is a general purpose detector capable of detecting and measuring the broadest range of particle signals. At the heart of the ATLAS detector lies the SemiConductor Tracker (SCT). It is a central part of the inner detector providing precision measurements of particle trajectories over a large $\eta$ range. The work presented in this thesis focuses on the performance and commissioning of the SCT detector. For Endcap-A, which was constructed at Nikhef, the assembly and calibration test results are described extensively in this work. The commissioning run in 2008 used cosmic rays to analyse and verify that the performance of the barrels was within specifications. The SCT uses silicon micro-strip technology, with the sensors produced by two different manufacturers: Hamamatsu and CiS. Hamamatsu produced all of the barrel sensor s and 82.8 % of the installed end-cap sensors. The remainder were produced by CiS, forming the middle/inner module types. The current-voltage (IV) scans of the sensors during module production showed a clear moisture dependence in different environments. Due to the different field plate geometries of the sensor manufacturer the CiS sensors behaved differently in a dry environment. The modules produced by the latter suffered from an earlier onset of micro-discharge and higher leakage currents. This was as a result of using the non-field plate strip configuration (where the metal strip is narrower than the width of the p-implant.) Fortunately, this effect could be easily controlled, by training the sensor. During 2004 and 2005 Nikhef constructed one of the SCT endcaps. Endcap-A has been extensively tested both at Nikhef during macro assembly and upon reception and integration at CERN. The endcap was successively installed, together with the TRT in its final position within the ATLAS cavern in May 2007. More than 98.7% of the optical links (used to communicate to and from the modules) were functional. The input noise to the amplifier of each module channel has been extensively tested from module production through to the final ATLAS installation. There were no significant changes in the noise even after all levels of transportation and integration. The grounding and shielding of the thermal enclosures as well as the connection to the TRT have proved highly efficient. After installation inside the ATLAS experiment 99.7% of the module channels are fully working, which is well within the specification of 1% allowed defective channels. The part of the detector made at Nikhef has only one dead and two problematic chips. A noise model has been developed and tested with measured noise results for thirteen SCT modules. A simulation of this model also gives a breakdown of the overall contributions to the noise. The most dominating of sources is the noise in the amplifier due to the base spread resistance. The total-strip capacitance is the dominating factor in the sensor. It is known that the total-strip capacitance changes as a function of bias voltage, so this has been investigated and implemented in the noise model. A fit has been made to the measured data points with the best fit outputting the behavioural form of the total-strip capacitance. A method of constraining the behaviour of the total capacitance has been investigated. Performing a fit on all SCT modules gives a good approximation of the geometrical values of Ctot: On initial application of the bias voltage, the total-strip capacitance takes a certain time before it will stabilise. This feature will not be a problem during the running of ATLAS, since the detectors will have been biased for a longer period of time. Knowing the total-strip capacitance, the depletion voltage parameter can be released f rom the noise model. This provides an extremely useful method for tracking the radiation damage of the detector and giving an overall estimate of the lifetime of the detector. The SCT Monitoring package is extremely important for monitoring all aspects of the module performance. The milestone 6 run during March 2008 was the first opportunity to test the entire reconstruction software chain, as well as providing preliminary results on the barrel detector performance and alignment. On September 10th, the LHC was officially switched on. During this time, events were recorded while a proton beam was circulating. For the remainder of 2008, the SCT continued to participate in global ATLAS cosmic runs.