Large Area Silicon Tracking Detectors with Fast Signal Readout for the Large Hadron Collider (LHC) at CERN

The Standard Model of elementary particles, which is summarized briefly in the second chapter, incorporates a number of successful theories to explain the nature and consistency of matter. However not all building blocks of this model could yet be tested by experiment. To confirm existing theories and to improve nowadays understanding of matter a new machine is currently being built at CERN, the Large Hadron Collider (LHC), described in the third chapter. LHC is a proton-proton collider which will reach unprecedented luminosities and center of mass energies. Five experiments are attached to it to give answers to questions like the existence of the Higgs meson, which allows to explain the mass content of matter, and the origin of CP-violation, which plays an important role in the baryogenesis of the universe. Supersymmetric theories, proposing a bosonic superpartner for each fermion and vice versa, will be tested. By colliding heavy ions, high energy and particle densities can be achieved and probed. This state of matter is called the quark gluon plasma and is similar to the state of our universe short after the Big Bang. In addition high precision measurments are performed to improve current models in high energy physics. The design of the two experiments related to this thesis are described in more detail in chapter~3: CMS, the 'Compact Muon Solenoid', which is an exploratory experiment symmetrically surrounding the interaction point and LHCb, the 'LHC beauty experiment', which is a forward spectrometer to CP-violation in the B-meson system. A crucial part of these experiments are the tracking detectors which measure the tracks of charged decay products after collision in the accelerator. From the bending under an applied magnetic field the momentum of charged particles can be derived. LHC imposes a harsh environment and heavy constraints on these detectors due to the fast readout requirements of 25\,ns as well as the harsh radiation environment. Silicon microstrip detectors are the favourable choice for large area tracking devices. The resulting radiation damage is well understood and it can be expected that they will withstand the accumulated dose of 10 years of operation at LHC. In addition signal formation occurs within the bunch crossing time of 25\,ns. Reproducibility at high precision is ensured by standard processes known from micro electronics industry. After a short introduction to semiconductor detectors in chapter~4 the layout and performance of the silicon trackers of the LHCb and CMS experiments are described in chapter~5. Radiation damage in silicon detectors under proton and neutron irradiation was investigated intensively within the ROSE collaboration (RD48). Based on their results good performance of the LHCb Inner Tracker is still expected after 10 years of operation. It was found that the degraded signal over noise ratio due to radiation damage of all modules lay above the given thresholds for track reconstruction efficiency and the expected depletion voltage after ten years of operation will be well below the specifications. Due to radiation damage the leakage currents inside the silicon detectors increase. This generates additional heat which in turn generates additional leakage current. If the provided cooling is not sufficient this could result in a break down called thermal runaway. Cooling for the sensors inside the detector boxes of the Inner Tracker is provided just via natural heat convection. A short review on natural heat convection and the equations used here are given in the appendix. It was investigated whether natural heat convection from the detector surface is sufficient to suppress thermal runaway. The results are summarized in chapter~5. In the Silicon Strip Tracker of the CMS experiment additional cooling to the sensors is provided via cooling pipes. In the outer part of the tracker the sensors are placed on a sophisticated rod structure which is later on inserted into a carbon barrel. An experimental setup to study the cooling performance of this rod structure is described in chapter~6. The temperature distributions on the fully equipped rod were measured under different conditions. For the first time the temperatures of the optohybrids when operated on the rod were measured. In a final test it could be concluded that the cooling performance of the rods is sufficient to guarantee a temperature difference of $\rm 10^o C$ between silicon surface and ambient. For understanding the behaviour of a detector it is essential to understand as well the behaviour of its readout electronics. The requirements for the readout electronics at LHC are quite demanding due to the high bunch crossing frequency. The readout chips for tracking detectors used for LHC show similar design principles: a charge sensitive preamplifier followed by a CR-RC shaper. In chapter~7 the BEETLE preamplifier chip used for the LHCb Silicon Tracker is introduced. Signal formation and potential noise sources are discussed. An in depth study of the resulting pulse shape is given, supported by numeric simulations. By means of Laplace transformation models for the pulse shape could be derived which were used later on for analysis. The parameter space of both the preamplifier and the shaper settings are investigated with respect to undershoot, spill-over and ballistic deficit. Performance studies on prototype modules under testbeam conditions are crucial for the final design of tracking detectors. Prototype modules were built for the LHCb Silicon Tracker with different length, thickness and readout pitch. To better understand the impact of strip width and readout pitch prototype modules from multigeometry sensors were built. The setup and the outcome of the comprehensive measurement program of the testbeam experiments are described in chapter~8. The main parameter describing the performance of silicon microstrip detectors is the ratio of signal over noise. This parameter was derived as both a function of the total strip capacitance and the impact point of the traversing particle. A signal loss is observed in the inter-strip region. Together with laboratory measurements of the strip capacitance the signal over noise ratio could be interpolated as a function of readout pitch and strip width. Bias voltage scans were performed to test whether the 500 mum thick detectors are sensitive to ballistic deficit. The studies were supported by numeric simulations of the signal formation in silicon sensors and suggest no significant signal loss due to ballistic deficit. In addition the measurements underline the hypothesis that the charge loss in the inter-strip region arrises due to charge traps on the sensor surface. Pulseshapes are studied for different bias voltage settings and shaper settings. The signal remainder after 25 ns of the peaking time contributes to the spill-over rate in the next event and is studied carefully. A parametrization as a function of the total strip capacitance could be given. Undershoot and crosstalk were investigated and tabled for the various prototype modules. The cluster finding efficiency is a crucial parameter for the tracking performance in the final experiment. It was found to be sufficient for all prototype modules, except the two- and three- sensor modules built from 320 mum multigeometry sensors. A direct comparison is given between the performance of a 500 mum thick three sensor modules once with long interconnect cable and once without. The interconnect cable is important for the Trigger Tracker station where the readout electronics is located outside the detector box. Before the detectors are placed into the final experiment they have to be tested carefully. Especially for long modules where silicon microstrip sensors are wire bonded together, short circuits and missing wire bonds are to be expected. It turned out that the internally generated testpulses on the readout chip are a powerfull tool to test for such defects. The response of the whole readout chip is highly dependent on the total capacitance on the input node of the preamplifier. Short circuits double the strip capacitance whereas missing wire bonds are leading to a decrease. Characteristic pulseshape parameters like the pulseheight are investigated whether the change due to the different capacitances are significant. Measurements were performed on a prototype module with artificially introduced defects. All defects could be detected at a significance higher than three standard deviations. No misidentification occured using an algorithm which automatically returns a channel map with type of defect and significance. Punch throughs in the dielectricum between implant strip and aluminum strip lead to a DC-coupling of the readout strips. In the case of the APV readout chip used for the CMS Silicon Strip Tracker it was shown that the occurance of about five such pinholes could lead to a loss of the whole readout chip. To detect pinholes a very sufficient method requiring no additional hardware is suggested based again on internal calibration pulses. The results of these tests together with a characterization of the testpulses are given in chapter~9.