Developing a Silicon Pixel Detector for the Next Generation LHCb Experiment

The second long shutdown of the LHC presents an opportunity for the LHCb experiment to upgrade its detector systems and switch to a fully software triggered readout. Its first tracking layer, the VELO detector, is no exception to this and is undergoing an upgrade increasing the number of sensitive channels from 180 thousand silicon microstrips to about 41 million pixels. The new system will operate with zero-suppressed readout at 40 MHz, while cooled down using evaporative liquid CO$_2$ in silicon microchannel plates. The VELO Upgrade will consist of 52 modules, placed around the beam-pipe, built at the University of Manchester and Nikhef. The construction of the modules is a complex process that consists of a number of tight tolerance steps, their results verified both in metrology and in the electrical and thermal performance testing. In order to store data and track the performance a database has been developed, used to automatically analyse the uploaded values as well as compute the grades and quality of the individual steps and final modules. By the end of August 2021, 42 modules have been produced in Manchester, 37 of them with high quality and no issues present. Due to the nature of the harsh radiation environment, the sensors have to withstand a fluence up to 1e16 1 MeV n$_\mathrm{eq} \mathrm{cm^{-2}}$ and still provide a good signal to noise ratio. A new method of a charge collection scan has been proposed, linking the commonly used voltage scan with a threshold scan and using the extrapolated tracking information to estimate the amount of collected charge. The simulation indicates that the scan of a subset of modules will take about 8 min, a feasible duration despite the impact on the physics data taking. A further upgrade of the LHCb is planned for Long Shutdown four of the LHC. This will operate at higher luminosities leading to a significant increase in the pile-up of the collisions from a single proton-proton bunch crossing. For this reason a precise time stamping $\mathcal{O}$(50 ps) is to be added. This could be achieved in silicon detectors by using $\mathcal{O}$(10) internal gain in the sensor. Simulations of the expected performance of a recently produced batch of sensors are presented. These characterise the anticipated performance of these $\mathcal{O}$(50 $\mu$m) segmented devices in a test beam, providing the impact of charge sharing and device response to an angular scan.