HOME: http://www.to.infn.it/wiki/doku.php?id=cms:xtrack
- Simulation of field and amplification
- Layout and manufacturing
- Testing field, charge collection
- ideas for testing, using radioactive system to test different part
The pixel and strip sensors currently used in particle physics experiments guarantee the desired space resolution. First of all, they tend to be thin (~100 micrometres), so one coordinate is given by the location of the sensor while the other two coordinates are well measured due to the size of the built-in segmentation (10–100 micrometres). To demonstrate how their timing characteristics can be improved to match those of advanced time detectors, we need to recall how charged particles generate a signal in a semiconductor sensor. The energy lost by a passing charged particle in the very thin wafer is sufficient to generate a large amount of negative (electrons) and positive (holes) charges, which are swept to the electrodes by the applied electric field and form the signal. The timing resolution depends upon the collection time of these charges; for example the entire signal from a 300 micrometres thick sensor can be collected in 3 nanoseconds. The timing resolution can be shortened in two ways: (i) increasing the speed of the electrons/holes by augmenting the electric field, and/or (ii) reducing the distance over which they have to drift, i.e., thinning the sensors. Both approaches have limitations: (i) the speed of electrons and holes saturates and cannot be increased beyond a (fairly high) limit, even with an increase in the electric field and (ii) thinning the sensor reduces the number of charge carriers contributing to the signal, making it too small for accurate measurements.
The challenge is therefore to reduce the collection time as much as possible while retaining a sufficiently large signal and controlling the effects of its fluctuations. The resolution of this conundrum is at the core of our idea: we propose to develop a thin silicon sensor with built-in charge multiplication. To this end, we want to exploit the impact ionization effect, which causes the generated charges to multiply while they are transported by the electric field. In so doing, the reduction of generated electron/hole pairs can be offset by a gain due to charge multiplication, resulting in the same signal strength as in non-thinned sensors. We foresee to develop sensors that are more than one order of magnitude thinner than the current design, with electric fields that can generate a gain of 10–100 without causing electrical breakdown. This goal can be achieved with an innovative design of the doping density within the sensor bulk. As the sensor operates with moderate gain and not in avalanche mode, it has very short dead time and it can reach very high counting rates (~1-10 GHz).
Using sensor models that we have developed over the years and results obtained within the RD50 collaboration, we have analysed how critical sensor parameters, such as capacitance, resistance and dark current, change in thin sensors and found that they retain an adequate signal-to-noise ratio. For example, the value of the capacitance remains dominated by the contributions from neighbouring pixels down to a thickness of about 2 micrometres.