Speaker
Description
Resistive Silicon Detectors (RSDs) are a recent innovation in the field of silicon sensors for 4D-tracking applications. The RSD design combines the LGAD technology with resistive read-out, yielding fast and large signals which are shared between multiple read-out pads. Thanks to their characteristics, RSDs can accurately reconstruct the hit position of an ionizing particle, achieving a spatial resolution equivalent to 4-5 % of their pitch. RSDs can thus provide the same spatial resolution as a traditional pixel detector with up to 100 fewer read-out channels: such an improvement leads to a drastic reduction in power consumption and cooling requirements, providing at the same time large pixels with plenty of space for the front-end electronics. On top of that, the RSDs feature an excellent timing resolution of 30-40 ps, making them very promising sensors for 4D-tracking applications.
This presentation will introduce, in particular, the latest evolution of the RSD design, the DC-coupled RSD (DC-RSD). In such a design, the read-out pads are DC-coupled to the resistive layer where the signal sharing happens, differently from the first RSD version (AC-RSD). In this improved design, the signal spreads over a well-defined area and is always seen by a predetermined number of read-out pads, and signals are short (1-2 ns). DC-RSDs, therefore, feature a very uniform spatial resolution and are suited for operation in high-rate environments.
After a brief introduction to DC-RSD, the talk will focus on recent test beam results obtained at DESY (5 GeV/c electrons) and CERN SPS (120 GeV/c hadrons) with DC-RSDs manufactured by FBK. Results from a variety of DC-RSD designs, featuring different geometries and pitches, will be presented, offering a thorough overview of the production and its performance. As an example, a 300-µm pitch DC-RSD achieved a spatial resolution of ~12 µm and a temporal resolution of ~40 ps.
A highlight of the talk will be the use of an innovative technique based on a Deep Neural Network (DNN) to reconstruct the position and timing of the particle hit, instead of the standard techniques, which rely on analytical sharing laws. The advantage is that the DNN takes the raw signals (i.e. a set of voltage values separated by a fixed time interval) as input, whereas the analytical laws make use of high-level variables (amplitude, area) which require a sophisticated offline analysis. It will be shown that, with the DNN approach, it is possible to envision a front-end electronics that reconstructs both the position and timing of the hit by sampling the DC-RSD signals multiple times. Such an advancement represents a key step towards the future development of a real 4D-tracker based on RSDs.
