Development and operation of tracking detectors in silicon technology for the LHCB upgrade

Resumo

This thesis covers most of my research from 2007 to 2013 as member of the High Energy Physics group in the Particle Physics department at the University of Santiago de Compostela. The LHCb experiment was the common topic during these years. At first, I was involved in the installation and commissioning of the Silicon Tracker detector. The ST is a silicon micro-strip detector which provides precise momentum measurements. The sensitive area is approximately 12 m^2 placed around the beam line, with a total of 272.600 electronic readout channels. Afterwards I joined the VELO upgrade project in the research and development of new sensor technologies for the LHCb upgrade. The VErtex LOcator is placed inside the beam pipe surrounding the LHCb interaction point, at only 7 mm from the beams, enclosed in a secondary vacuum box. The role of the VELO is critical in the overall performance of LHCb, providing excellent vertex and impact parameter resolution, high efficiency, and fast pattern recognition for triggering purposes. At the beginning of 2007 the production of the Silicon Tracker modules was finished and I got involved in the detector installation in the LHCb cavern. For the next three years I worked in the design and development of the detector control system (DCS). At the same time I helped in the detector commissioning with LHC particle beams. These activities allowed me to gain a detailed knowledge of the detector, as the DCS controls all the aspects related to the detector performance. The DCS tasks involves the power-up, sensor biasing, control and configuration of the electronics and monitoring of the environmental variables to ensure safe operation of the detector. The LHCb's IT team provided guidelines for the experiment control system that we should follow in order to allow a proper integration and synchronization with the experiment. However we found some additional issues related to the detector safety that were not covered. For that reason we implemented a parallel control tree called Safety Tree which look after the detector status and take automatic actions if a problem is detected. The Safety Tree is made of a set of hierarchical rules which evaluates the nature of the problem and decides if is necessary to shut down, in a controlled way, the smallest partition possible. Thus the Safety Tree anticipates the DSS actions, which would take a much more sudden shut-down of the whole detector. We also implemented alarm services that send emails and SMS to the experts responsible of the detector safety. The control system was completed in time for the LHCb commissioning in 2009, but the work continued in the code maintenance and improvement one more year, until Sandra Saornil took the responsibility of the project. In 2010 I joined the VELO upgrade project. The upgraded VELO will be installed in 2018. It will be placed closest to beam, at 5.1 mm instead of the current 7 mm. Therefore it will have to cope radiation conditions and data bandwidth much more demanding than the current ones. For the upgraded VELO sensors two possible technologies were under consideration: silicon micro-strips and pixels. The first option is an evolution of the current VELO sensors, taking advantage of technological progress that have appeared since the construction of the detector. For this option, I was responsible for the assembly and characterization of prototypes to address different issues of the micro-strip technology. The first micro-strip prototype in which I was involved was built in 2010. It was made of a micro-strip sensor, called PR01, wire bonded to a Silicon Tracker hybrid. The PR01 sensor was manufactured by Hamamatsu on a n+-on-n basis. It follows a radial geometry with quasi-circular shape and an aperture of 72o. The sensor has two different regions: the innermost one has a pitch of 40 microns, while the outermost one has a pitch of 60 microns. I carried out the prototype characterization with a 120 GeV/c pion beam in the CERN SPS North area facilities. One of my responsibilities was to develop the analysis code for the PR01. Such code was used later to analyse all the micro-strip sensors that were tested in the following test-beam campaigns, and it was integrated into the larger test-beam software package. The data collected during the test-beam showed that the sensor could achieve a resolution of 5.6 microns for particles at the optimum angle of ~ 7º In 2011 I worked in the assembly and test of a new sensor, the D0, which is a p+-on-n type with parallel micro-strips manufactured by Hamamatsu. The D0 sensor has the particularity of including intermediate strips that are not instrumented, i.e. not connected to the readout electronics. In this case the pitch was 30 microns, although being instrumented alternately, so the distance between readout strips is 60 microns. The resolution achieved with the D0 during the test- beam at the SPS facility was 9.5 microns. This is a much better resolution than the theoretical (binary) 17.6 microns. The prototype efficiency is greater than 99.5%. During the construction phase of the D0 prototype I developed along with Eliseo Perez and Francisco Rey a novel process for manufacturing pitch adapters by laser ablation. This procedure proved to be very flexible and robust, and allow us to design, fabricate, test and implement improvements in a single day. Thanks to it, we successfully produced pitch adapters for the D0 sensor and many other micro-strip prototypes. With the accumulated knowledge and experience of these and other sensors, new micro-strip sensors were designed in order to withstand the radiation of the LHCb with high luminosity while keeping, if not improving, the performance of the current VELO. Such sensors were manufactured by Hamamatsu with an R and Phi geometries, and delivered in mid-2012. The sensors were built on silicon wafers of 150 and 200 ¿m thickness. They are n+-on-p type, and the pitch increases continuously from the inner region (30 microns) to the periphery (114 microns). Once the sensors were received in our facilities at Santiago, together with Eliseo, I proceeded with the electrical characterization to ensure that they achieve the manufacturing requirements. I developed a procedure to check that the metrology of the sensors met the specifications. I found out that the warp of the 150 microns thick sensors is 50% higher than their counterparts in 200 microns, as expected. Finally we built a prototype with a radial sensor, which was characterized with two laser beams of 660 nm and 1060 nm and a beta source of 2 MeV betas (90Sr). The sensor was instrumented with a readout hybrid developed for the Silicon Tracker, with a new pitch adapter developed by the procedure mentioned above, and read out by the Alibava system. After developing the analysis code it was demonstrated that the sensor signal to noise ratio is above of 30, and it is kept constant throughout the sensor, and independent of the pitch. In addition, I was involved in the pixel technology within the VELO upgrade. In late 2012 we tested in the SPS area a pixel sensor bump-bonded to a Timepix chip, and two sensors bump-bonded to Medipix3 chips. These Medipix3 assemblies were designed in Santiago and manufactured by CNM-Barcelona. Afterwards they were irradiated to a dose of 2.5 × 1015 1MeVneq/cm2 at Ljubljana Neutron Irradiation Facility. The purpose was to analyse different guard ring structures and test two different thickness of the sensor. We need to ensure that they can withstand full bias voltage after irradiation levels similar to the expected for the VELO upgrade conditions. The irradiated sensors were identical but the distance from the last pixel to the edge of the sensor. In one case it was 250 microns and in the other it was 400 microns. The test-beam data showed that the efficiency could reach 94% when we discard charge sharing issues (particle hitting the centre of the pixel), and fell to 82% when we consider all the particles hitting the assembly. The sensor with less distance between the last pixels to the edge of the sensor had a efficiency below 20%. These results, among others and the result of simulations, were part of the evidence supporting the decision taken in June of 2013 to choose 200 microns thick pixel sensors for the future VELO for the LHCb with high luminosity.