Overview of the ALICE ITS3 Upgrade

The ALICE ITS3 upgrade replaces the innermost tracking layers with a fully cylindrical, self-supporting silicon vertex detector using 65nm CMOS Monolithic Active Pixel Sensors and wafer-scale stitching to achieve an ultra-low material budget of less than 0.09% X$_0$ per layer while transitioning to air cooling.

The Gist

Imagine the ALICE experiment as a high-speed camera trying to take pictures of tiny, fleeting particles created when protons smash together. To get a clear picture, the camera needs a lens that is incredibly close to the action but doesn't get in the way.

The paper describes a massive upgrade to this "lens," called ITS3, which is essentially a new, ultra-thin skin for the detector. Here is how it works, broken down into simple concepts:

1. The Problem: The Old Lens Was Too Clunky

The previous version (ITS2) was like a heavy, bulky winter coat made of many layers. It had:

• Sturdy frames: Rigid supports to hold the sensors.
• Thick wiring: Lots of cables and circuit boards (like flexible printed circuits) to carry power and data.
• Water pipes: A complex plumbing system to cool the sensors down because they got hot.

All this extra stuff (the coat, the wires, the pipes) got in the way of the particles, making it harder to track them accurately, especially the very short-lived ones.

2. The Solution: A "Bent-Wafer" Skin

The new ITS3 upgrade is like replacing that heavy winter coat with a single, ultra-thin, flexible sheet of silk.

• The "Silk" (The Sensors): The team made the silicon sensors incredibly thin (50 micrometers—thinner than a human hair). Because they are so thin, they can be physically bent into a cylinder shape, hugging the beam pipe tightly.
• No More Frames: Because the silicon is strong enough on its own when bent, they don't need the heavy metal frames or support structures anymore. It's a self-supporting structure.
• The "Seamless" Stitch: To make these sensors long enough to cover the whole cylinder (about 26 cm), they had to stitch multiple pieces of silicon together. Imagine sewing two pieces of fabric together so perfectly that you can't tell where the seam is. They did this at the microscopic level, creating one giant, seamless sensor.

3. The "Smart" Chip: Integrating the Electronics

In the old design, the "brain" (electronics) was separate from the "eye" (sensor), requiring thick wires to connect them.

• The Upgrade: By using a newer, smaller manufacturing process (65 nm), they built the power and data electronics directly onto the silicon sensor itself.
• The Result: It's like having the camera's battery and processor built right into the lens glass. This eliminates the need for bulky external wires and circuit boards, saving a huge amount of space and weight.

4. Cooling: From Water Pipes to a Gentle Breeze

The old system needed water pipes to cool the sensors, which added even more weight.

• The New Way: The new sensors use so little power that they don't need water. Instead, they use air cooling.
• The Analogy: Think of it like a computer fan blowing air over a laptop. They use special, ultra-lightweight foam (like a sponge made of carbon) that acts as a heat exchanger. Air blows over this foam, carrying the heat away. Tests showed that a gentle breeze (about 5 meters per second) is enough to keep the sensors cool and stable, without causing them to vibrate.

5. The Proof: Testing the Prototype

Before building the final version, the team built test models (called MOSS and MOSAIX) to make sure the "stitching" and "bending" would work.

• The Stitch Test: They successfully stitched sensors together to make long, continuous sheets.
• The Results: The tests were a huge success. The sensors worked with a 98% success rate (very few defects). They proved that the sensors could detect particles with high precision (better than 5 micrometers) and that the air cooling kept them stable without shaking the image.

The Bottom Line

By switching to this new design, the ALICE experiment is reducing the "material budget" (the amount of stuff the particles have to pass through) by 75% (from 0.36% down to 0.09%).

In simple terms: They replaced a heavy, water-cooled, wire-filled camera lens with a feather-light, air-cooled, seamless skin. This allows the camera to see the tiniest, fastest particles much more clearly than ever before.

Technical Summary

Technical Summary: Overview of the ALICE ITS3 Upgrade

Problem Statement
The ALICE experiment aims to enhance its capability to track short-lived particles, such as charm and bottom quarks, particularly at low momenta. While the current Inner Tracking System (ITS2), installed during Long Shutdown 2, performs well, its material budget limits the detector's pointing resolution. The ITS2 inner barrel relies on stave support structures, flexible printed circuit boards (FPCBs) for data and power transmission, and water-cooled plates for thermal management. These components contribute significantly to the material budget (0.36% $X_0$ per layer), causing multiple scattering that degrades tracking precision. To prepare for the high-luminosity environment of LHC Run 3 and beyond, a drastic reduction in material is required without compromising detector performance.

Methodology and Design
The proposed solution is the Inner Tracking System 3 (ITS3), a next-generation vertex detector scheduled for installation during Long Shutdown 3 (2026–2030). The ITS3 design fundamentally shifts from modular silicon assemblies to a "bent-wafer" architecture utilizing Monolithic Active Pixel Sensors (MAPS) fabricated in a 65 nm CMOS process. The methodology relies on three primary innovations:

1. Structural Geometry and Self-Support: Sensors are thinned to 50 µm, leveraging the inherent flexibility of silicon to bend them into semi-cylindrical shapes around the beam pipe. This creates a self-supporting structure that eliminates the need for traditional stave supports, retaining only minimal carbon foam at the edges. The design features six half-layers with radii ranging from 19 mm to 31 mm.
2. Wafer-Scale Stitching and Circuit Integration: Moving from the 180 nm process (used in ITS2) to 65 nm allows for higher transistor density and the integration of power and signal electronics directly onto the silicon chips. Because the required sensor length (266 mm) exceeds standard lithographic reticle sizes, a "stitching" technique is employed. This connects multiple design units electrically during fabrication, creating seamless, 27 cm-long sensors with integrated power distribution and readout. This eliminates the need for external FPCBs within the active volume.
3. Air Cooling: The system replaces water cooling with air convection. Thermal management is achieved using two types of carbon foam: open-cell foam rings (RVC Duocel) in direct contact with sensors to act as heat exchangers, and dense, graphite-infused foam (Allcomp K9) along the edges to maintain cylindrical geometry.

Prototype Validation (MOSS and MOSAIX)
The R&D program has been validated through two full-scale prototypes:

• MOSS (MOnolithic Stitched Sensor): Developed during Engineering Run 1 (ER1), this prototype demonstrated the feasibility of the stitching process. It comprises 10 Repeated Sensor Units (RSUs) with independent power and readout capabilities. Testing involved 82 sensors to evaluate yield, reliability, and beam performance.
• MOSAIX (MOnolithic Stitched Active pIXel): Developed for Engineering Run 2 (ER2), this sensor serves as the functional prototype for the final design. It features a more integrated architecture where power and control signals are distributed through the Left and Right Endcaps (LEC/REC). It includes 12 RSUs, each subdivided into 12 independent tiles, allowing for selective power switching and testing of 12 distinct pixel geometries to finalize the optimal design for production.

Key Results

• Material Budget: The integration of self-supporting sensors, on-chip electronics, and air cooling reduces the material budget from 0.36% $X_0$ (ITS2) to approximately 0.09% $X_0$ per layer. The material distribution is nearly uniform, with only minimal glue and ultra-lightweight carbon fiber supports within the active volume.
• Yield and Reliability: The MOSS prototype achieved a laboratory yield of approximately 98%, excluding metal stack defects caused by manufacturing imperfections which affected about 14% of RSUs. These defects were reported to the foundry to improve future production.
• Performance Metrics: Beam tests confirmed a detection efficiency above 99% and a fake-hit rate below $10^{-1}$ hits/pixel/s. Measured spatial resolutions ranged from 4 to 5.5 µm across different pixel pitches, validating that an intermediate pixel size can meet the 5 µm target.
• Thermal and Mechanical Stability: Wind tunnel tests on a breadboard model indicated that a freestream air velocity greater than 5 m/s is sufficient to maintain a temperature variation of less than 10 K across the pixel matrix, given a power density of ~40 mW/cm². Vibration analysis showed that at an airspeed of 8 m/s, maximum displacement is less than 0.5 µm, well within the 2 µm design limit.

Significance
The paper claims that the ITS3 upgrade represents a fundamental shift in vertex detector technology. By successfully validating wafer-scale stitching, high-yield production of thinned sensors, and air-convection cooling, the collaboration has demonstrated a path to a detector that is virtually free of non-active material. The transition to the 65 nm process and the elimination of external infrastructure (FPCBs, water cooling, support frames) are critical for achieving the necessary pointing resolution for future heavy-ion physics analyses. The progression from the MOSS prototype, which confirmed stitching feasibility, to the MOSAIX prototype, which validates integrated data/power distribution and finalizes pixel geometry, establishes the technical readiness for the final qualification model and subsequent production.