Beam-test evaluation of pre-production Low Gain Avalanche Detectors for the ATLAS High Granularity Timing Detector

This paper presents beam-test results demonstrating that pre-production Low Gain Avalanche Detectors for the ATLAS High Granularity Timing Detector meet all performance requirements, including charge collection, time resolution, and hit efficiency, even after neutron irradiation simulating end-of-life High Luminosity-LHC conditions.

The Gist

Imagine the Large Hadron Collider (LHC) as the world's most energetic particle accelerator, smashing protons together to recreate conditions just after the Big Bang. As scientists upgrade this machine to the "High Luminosity" phase, they are essentially turning up the volume on the noise. Instead of a few particles passing through at a time, they will be bombarding the detectors with a blizzard of collisions happening all at once. This "pile-up" makes it incredibly hard to tell which particle came from which collision.

To solve this, the ATLAS experiment is building a new, ultra-fast camera called the High Granularity Timing Detector (HGTD). Think of this detector not just as a camera that takes pictures, but as a high-speed video camera that can freeze time so precisely it can distinguish between two events happening a billionth of a second apart.

The heart of this new camera is a special type of silicon sensor called a Low Gain Avalanche Detector (LGAD). You can think of an LGAD as a "smart microphone" for particles. When a particle hits it, the sensor doesn't just hear a whisper; it amplifies the signal so it can be heard clearly, even in a noisy room.

The Stress Test: Simulating a Harsh Environment

The paper describes a rigorous "stress test" these sensors underwent before being approved for the final camera. The environment inside the LHC is brutal; it's like a nuclear reactor where sensors are constantly bombarded by radiation. Over time, this radiation damages the sensors, much like how constant sunlight fades a painting or rust eats away at metal.

To prepare for this, scientists took pre-production sensors and subjected them to a "radiation bath" at a nuclear reactor in Slovenia. They blasted them with neutrons until they had absorbed as much radiation as they would see over the entire lifetime of the upgraded LHC (up to 2.5 × 10¹⁵ neutrons per square centimeter). It's like taking a new car, driving it through a sandstorm for a million miles, and then checking if the engine still runs.

The Results: Do They Still Work?

The team tested these "beaten-up" sensors at two major particle physics labs (CERN in Switzerland and DESY in Germany) using high-speed particle beams. They looked at three main things:

1. The Signal (Charge Collection):

   • The Goal: The sensor needs to catch enough "electric charge" from a passing particle to be useful.
   • The Result: Even after being blasted with maximum radiation, the sensors still collected enough charge to work. Interestingly, the paper found that if the particle hits the sensor at a slight angle (like a raindrop hitting a windshield rather than falling straight down), the sensor actually collects more charge. This is because the particle travels a longer path through the sensor, leaving a bigger trail of energy.

2. The Speed (Time Resolution):

   • The Goal: The sensor needs to time the arrival of a particle with extreme precision (better than 50 picoseconds, which is 50 trillionths of a second).
   • The Result: The sensors passed this test with flying colors. Even the most damaged sensors could time events with the required precision, provided they were given a little extra electrical "push" (voltage) to overcome the radiation damage.

3. The Reliability (Efficiency):

   • The Goal: The sensor needs to detect almost every particle that passes through it (at least 95% of the time).
   • The Result: The sensors were incredibly reliable. They detected particles with over 99% efficiency when new, and still maintained over 95% efficiency even after the heavy radiation damage. The tests showed that the sensors work uniformly across their entire surface, meaning no "dead spots" appeared after the stress test.

The Verdict

The paper concludes that these specific sensors, made by two different teams (IHEP and USTC in China), are ready for the job. They proved they can survive the harsh, radiation-filled environment of the future LHC while still acting as ultra-fast, precise timers.

In short, the scientists built a prototype "smart microphone," threw it into a hurricane of radiation, and found that it still hears every whisper perfectly. This gives them the confidence to install millions of these sensors into the ATLAS detector, ensuring they can untangle the complex web of particle collisions in the future.

Technical Summary

Technical Summary: Beam-test evaluation of pre-production Low Gain Avalanche Detectors for the ATLAS High Granularity Timing Detector

Problem Statement
The High Granularity Timing Detector (HGTD) is a critical component of the Phase-II upgrade for the ATLAS experiment at the High-Luminosity Large Hadron Collider (HL-LHC). Its primary function is to mitigate pile-up effects in the forward region ($2.4 < |\eta| < 4.0$) and enable per-bunch luminosity measurements by combining precise timing with spatial tracking. The detector relies on Low Gain Avalanche Detectors (LGADs), which must operate under extreme radiation conditions, specifically up to a 1 MeV neutron-equivalent fluence of $2.5 \times 10^{15} \, \text{neq/cm}^2$.

The challenge lies in ensuring that pre-production LGAD sensors, manufactured with carbon-enriched gain layers to improve radiation hardness, meet stringent performance targets throughout their operational lifetime. These targets include a time resolution of 40 ps (unirradiated) and 50 ps (irradiated), a collected charge greater than 15 fC (unirradiated) and 4 fC (irradiated), and hit reconstruction efficiencies exceeding 97% (unirradiated) and 95% (irradiated) in the central pad area. Validating these sensors prior to mass production is essential to confirm their suitability for the HL-LHC environment.

Methodology
The study evaluated pre-production LGAD sensors from two design groups: the Institute of High Energy Physics (IHEP) and the University of Science and Technology of China (USTC), both manufactured by the Institute of Microelectronics (IME). The sensors featured a 50 $\mu$m active thickness and a 775 $\mu$m total thickness (with some pre-production variants at 300 $\mu$m).

The evaluation involved a comprehensive beam-test campaign conducted at CERN (using a 120 GeV pion beam) and DESY (using a 5 GeV electron beam) in 2023 and 2024. The experimental setup included:

• Reference Systems: A six-plane MIMOSA pixel telescope for track reconstruction and a Micro-Channel Plate Photomultiplier Tube (MCP-PMT) for time reference (calibrated to $\sim$10.6 ps resolution).
• Device Under Test (DUT): Single-pad LGADs mounted on custom readout boards with on-board amplification, cooled to approximately -30°C.
• Irradiation: Sensors were exposed to fast neutrons at the TRIGA reactor in Ljubljana, Slovenia, to fluences of 0, 0.8, 1.5, and $2.5 \times 10^{15} \, \text{neq/cm}^2$.
• Data Acquisition: Waveforms were captured via a 1 GHz oscilloscope, and triggers were managed by a FE-I4 chip and a Trigger Logic Unit (TLU) within the EUDAQ2 framework.
• Analysis: Data processing involved waveform digitization, pedestal subtraction, and Constant Fraction Discriminator (CFD) time extraction. Track reconstruction was performed using the Corryvreckan framework. Performance metrics were analyzed as a function of bias voltage and beam incident angle (0°, 6°, 12°, 18°).

Key Contributions and Results
The paper presents a detailed characterization of the sensors' charge collection, time resolution, and hit efficiency under varying radiation doses and operational conditions.

1. Collected Charge:

   • Unirradiated sensors consistently collected charges exceeding 15 fC even at low bias voltages.
   • Irradiated sensors showed reduced charge collection due to acceptor removal in the gain layer. However, at the maximum fluence of $2.5 \times 10^{15} \, \text{neq/cm}^2$, all sensors achieved the required >4 fC threshold at bias voltages below 550 V (the safe operating limit).
   • A significant increase in collected charge was observed with increasing incident angle (up to 18°). This is attributed to a longer particle path length through the sensor and a reduction in charge screening effects on the gain layer, particularly for unirradiated sensors.

2. Time Resolution:

   • Time resolution was derived by subtracting the MCP-PMT contribution from the residual distribution between the DUT and the reference.
   • Unirradiated sensors achieved time resolutions below 40 ps.
   • Irradiated sensors maintained time resolutions below 50 ps across the tested fluences, with performance generally improving or stabilizing at higher bias voltages, provided the voltage remained below the breakdown threshold.
   • Time resolution showed minimal dependence on the incident beam angle, with relative variations remaining within a few percent.

3. Hit Reconstruction Efficiency:

   • For unirradiated and moderately irradiated sensors, hit efficiency remained above 99% across the tested bias range.
   • Even at the highest fluence ($2.5 \times 10^{15} \, \text{neq/cm}^2$), all sensor designs met the minimum 95% efficiency requirement at appropriate bias voltages.
   • Spatial uniformity studies revealed that the active area remained consistent across the sensor surface. Interestingly, irradiated sensors exhibited a slight apparent increase in the active area (defined by the 50% efficiency point), likely due to radiation-induced modifications in the electric field distribution rather than physical expansion.

Significance and Claims
The authors conclude that the pre-production LGAD sensors from both IHEP and USTC designs successfully meet the HGTD performance requirements for charge collection, time resolution, and hit efficiency, even under the extreme radiation conditions expected at the end of the HL-LHC lifetime.

The study validates the radiation hardness of the carbon-enriched gain layer design, demonstrating that the sensors can operate within safe bias voltage limits (below 550 V) while maintaining the necessary signal quality. The results support the deployment of these specific sensor designs in the final HGTD detector. The paper notes that while these single-pad tests are promising, further validation using full-size sensors hybridized with the final ALTIROC readout ASICs and comprehensive system-level tests are ongoing to confirm the feasibility of the complete timing detector system.