Speaker
Description
The ALICE Inner Tracking System upgrade (ITS3) will employ stitched, wafer-scale Monolithic Active Pixel Sensors (MAPS) for the first time in a high-energy physics detector. The first stitched prototype, the Monolithic Stitched Sensor (MOSS), underwent testing that confirmed yield compliance with ITS3 requirements. Linearity between time-over-threshold and deposited energy was validated from 1.8 to 5.9 keV using $^{55}$Fe X-rays. In-beam tests confirmed the device meets ITS3 efficiency (99%), and fake-hit rate targets (< 10$^{-6}$ hits pix$^{-1}$ events$^{-1}$), with performance sustained up to expected irradiation levels (400 krad, 4$\cdot$10$^{12}$ 1 MeV n$_{eq}$). This talk presents validation steps and characterisation results.
Summary (500 words)
The upgrade of the ALICE Inner Tracking System 3 (ITS3) reduces the material budget of its three innermost layers from 0.35% X0 to about 0.07% X0 per layer. The remaining contribution is coming almost solely from the sensors themselves. This is achieved by employing wafer-scale Monolithic Active Pixel Sensors (MAPS) with a size of up to 265 mm x 98 mm, bent into half-layers and held in place by low-density carbon foam. The targeted power consumption of 40 mW/cm2 allows for the use of air instead of water cooling.
The Monolithic Stitched Sensor (MOSS) is the first large-scale prototype covering the full length of the ITS3. The sensor consists of 10 Repeated Sensor Units (RSUs), containing the pixel matrices subdivided into eight regions. The regions prototype two different pixel pitches and six variants of the in-pixel front end. The RSUs are connected by stitching to the power supply and readout circuitry.
Due to the large size of the ITS3 sensors, single points of failure could compromise an entire half-layer. To avoid this scenario, the final sensor design will allow faulty regions to be individually disconnected from the global power net. In a series of powering and functional tests, the yield of the MOSS prototype was assessed. The functionality results per tested wafer are shown in fig. 1, revealing significant wafer-to-wafer variation. The yield was affected by non-powerable units due to production-related shorts in the power distribution metal layers. Also, shortcomings of the prototype readout architecture influenced the yield. Excluding the power metal faults issues and prototype readout issues, approximately 97% of regions pass the functional tests.
In laboratory measurements, the operation point for each frontend implementation was optimised. The required efficiency of > 99% maintaining a fake-hit rate of < 10-6 hits pix-1 events-1 was demonstrated in in-beam test campaigns at CERN’s PS and SPS. The robustness of the required performance was validated up to a non-ionising dose of 4 x 1012 1 MeV neq and an ionising dose of 400 krad (fig. 2.). In-beam performance differences between the frontend variants are compatible with chip-to-chip variations.
As the Time-over-Threshold (ToT) of a signal on MOSS is proportional to the deposited energy in a pixel, its measurement can provide insights into the charge-collection properties of the sensor. However, the ToT is subject to inter-pixel variations smearing out the spectra. This spread can be mitigated by measuring the ToT as a function of the deposited energy for each pixel. Illuminating the MOSS with X-ray emissions produced by 55Fe, a ToT-spectrum was measured (fig. 3). Relating the peak positions with the emission energies, the linearity of the frontend response between the 1.8 and 5.9 keV was demonstrated.
In summary, the MOSS validates the use of stitched sensors in terms of yield and performance for use in ITS3 and beyond. The tape-out of the final full-scale prototype, the MOSAIX, is currently under preparation.
