Exploring unique design features of the Monolithic Stitched Sensor with Timing (MOST): yield, powering, timing, and sensor reverse bias
H. Wennloef · 2025
Read Exploring unique design features of the Monolithic Stitched Sensor with Timing (MOST): yield, powering, timing, and sensor reverse bias by H. Wennloef (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-13696.txt.
Introduction
A large sensor team reports first tests of a stitched timing chip in this 2025 paper. The chip serves a planned upgrade of a major collider tracker and related research. Unlike its sibling design, it uses one shared power domain with switches that can cut off faulty patches.
Hits are sent out at once, keeping their timing, rather than on a fixed clock. Bias is fed through the front-end electronics instead of the chip back. The paper reports yield counts, fine timing reads, and bias trials. The authors judge how these odd design choices help or hurt sensor performance.
Catalog record
| CREATOR | H. Wennloef |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 12 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-13696
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Exploring unique design features of the Monolithic Stitched Sensor with Timing (MOST): yield, powering, timing, and sensor reverse bias H. Wennloef Abstract Monolithic stitched CMOS sensors are explored for the upgrade of Inner Tracking System of the ALICE experiment (ITS3) and the R&D of the CERN Experimental Physics Department. To learn about stitching, two 26 cm long stitched sensors, the Monolithic Stitched Sensor (MOSS), and the Monolithic Stitched Sensor with Timing (MOST), were implemented in the Engineering Round 1 (ER1) in the TPSCo 65nm ISC technology. Contrary to the MOSS, powered by 20 distinct power domains accessible from separate pads, the MOST has one global analog and digital power domain to or from which small fractions of the matrix can be connected or disconnected by conservatively designed power switches to prevent shorts or defects from affecting the full chip. Instead of the synchronous readout in the MOSS, the MOST immediately transfers hit information upon a hit, preserving timing information. The sensor reverse bias is also applied through the bias of the front-end rather than by a reverse substrate bias. This paper presents the first characterization results of the MOST, with the focus on its specific characteristics, including yield analysis, precise timing measurements, and the potential of its alternative biasing approach for improved sensor performance. 1 Now with University of Geneva, Switzerland. 2 Now with Brookhaven National Laboratory, US. 3 Now with Miromico IC, Zurich, Switzerland. 1 Introduction During LS3 (2026–2030), the ALICE Inner Tracking System (ITS) will undergo a major upgrade with the replacement of its three innermost layers as part of the ITS3 project [ 1 ] . This upgrade aims to enhance vertexing precision, increase physics yield, and improve overall detector performance in preparation for the High Luminosity LHC (HL-LHC) . The ITS3 project aims to position the three inner layers of the detector closer to the interaction point, with the first layer at 19 mm, while reducing the material budget to 0.09% X_{0} per layer. This is achieved by employing a wafer-scale, stitched monolithic sensor chip that integrates all electrical functions directly on-chip. The chip is thinned to 50 \upmu m and bent around the beam pipe, creating a vertex detector composed of cylindrical layers formed almost entirely by the silicon sensors themselves. The TPSCo 65 nm CMOS imaging technology has been qualified for high-energy physics [ 2 , 3 , 4 ] in the framework of the R&D of the CERN Experimental Physics department, and the ALICE ITS3 upgrade project. Engineering Round 1 (ER1) includes two wafer-scale, 26 cm long, stitched sensors, the Monolithic Stitched Sensor (MOSS) [ 5 , 6 ] and the Monolithic Stitched Sensor with Timing (MOST). The MOST contains 901120 pixels distributed across 10 stitched Repeated Sensor Units (RSUs) of 25.5 mm by 2.5 mm each, designed with a pixel pitch of 18 \upmu m \times 18 \upmu m. The MOST design explores a highly granular powering scheme to address yield, an asynchronous readout scheme offering timing capabilities, and an alternative approach to reverse sensor biasing. This contribution describes the initial measurement results of the MOST, focusing on these specific characteristics, to present an outlook for future advancements in CMOS monolithic sensors. 2 Highly granular powering scheme using power switches and dense matrix circuitry Contrary to the MOSS, segmented in 20 fully separated power domains accessible by separate pads, the MOST is equipped with a much more granular powering scheme: power switches, fully programmable through the slow control interface of the chip, can connect or disconnect small portions of the matrix to or from the global, conservatively designed power network. The aim is to isolate shorts or defects and prevent them from affecting the full chip, allowing the internal circuitry powered through the power switches to be designed at the full density, so at the maximum density defined by the design rules. Figure 1 : MOST Layout of a group of 4 pixels [ 7 ] . Figure 1 shows the layout of a group of 4 pixels, and illustrates the conservative design of the power switches, and the dense design of the rest of the circuitry. Figure 2 : Graph on the left showing the increase in power consumption of the chip with the number of pixels powered on. On the right there is first evaluation of the individual pixel yield for 4 chips or 40 stitched units (SU): the average number of non-fully responsive pixels per stitched unit is 15 per 90112 pixels, indicating a pixel yield better than 99.98%. The digital circuitry of the MOST is power gated along the pixel columns in subsets of 352 pixels, resulting in a total of 2,560 independently controlled digital switches. The analog power is grouped per 256 pixels, arranged in 4 rows of 64 pixels, amounting to 3,520 analog switches. This difference in segmentation of analog and digital power resulted in some design complication, and has been avoided in the MOSAIX, the full-size, fully functional prototype for the ITS3 upgrade, which adopts the same principle of power switches and global power supplies with reduced granularity. To evaluate the validity of this powering approach, power-on tests were performed first. A processing issue, expected to be resolved in the next fabrication, caused shorts on the global power network for approximately 30% of the chips measured so far. This issue also affected the MOSS and was previously reported in Ref. [ 8 ] . The 30% failure rate is consistent with observations on the MOST chip, where a significant overlap in the global power network between two metals prone to shorts has been identified. All MOSTs which could be powered on with all power switches in the matrix turned off, were found to be functional, and allowed gradual powering of the full chip using the power switches. The power consumption measurements presented in Figure 2 (left) exhibit linear increase in power consumption as more areas are activated by the power switches, and confirms the effectiveness of this fine-granularity power distribution approach in managing power. As a further validation, particularly regarding the feasibility of designing the pixel circuitry at the full density, the detailed pixel yield was evaluated on four different chips. Figure 2 (right) shows the number of non-responsive pixels per stitched unit or per 90112 pixels for these 4 chips, i.e. for 40 stitched units. The average number is 15 pixels per 90112 indicating a pixel yield above 99.98%. The 160 non-fully responsive pixels forming an outlier stitched unit, are actually within the same 3 or 4 rows, which hints at a defect in one of the shared lines within those rows. This one outlier increases the average non-fully responsive pixels per RSU from 11 to 15. While this yield analysis is done with limited statistics, the excellent individual pixel yield demonstrates the feasibility of high-density designs for future applications. 3 Timing Capabilities with Asynchronous Readout Figure 3 : Figure on the left shows the schematic of signal propagation from the pixel to the chip output. Figure on the right illustrates how hit information is encoded within the output waveform, including the address of the RSU (SID), Matrix ID (MID), Unit ID (UID), Pixel ID (PID), and Column ID (CID). The asynchronous readout in