Studying AC-LGAD strip sensors from laser and testbeam measurements

This paper demonstrates that a calibrated 1060 nm laser setup can effectively characterize the spatial and timing resolutions of AC-LGAD strip sensors with results compatible to 120 GeV proton testbeam measurements, offering a valuable tool for accelerating R&D in 4D tracking detectors supported by TCAD and Weightfield2 simulations.

The Gist

Imagine you are trying to catch a speeding bullet with a net. To do this effectively, you need two things: you need to know exactly where the bullet hit your net, and you need to know exactly when it happened, down to a trillionth of a second.

This paper is about building a better net for catching subatomic particles in high-energy physics experiments (like those at the Large Hadron Collider). The "net" is a special type of silicon sensor called an AC-LGAD.

Here is the story of how the researchers tested these sensors, explained simply.

1. The Problem: The "Pixel" vs. The "Strip"

Traditional sensors are like a grid of tiny, separate buckets (pixels). If a particle lands right on the line between two buckets, the signal gets messy, and you lose precision. Also, the walls between the buckets take up space, meaning you can't catch particles in 100% of the area.

The new AC-LGAD sensors are different. Instead of separate buckets, imagine a long, continuous trampoline (a resistive strip). When a particle hits anywhere on the trampoline, the bounce travels to the edges. By measuring how the "bounce" is shared between the two edges, the computer can figure out exactly where the hit happened. This gives them a "100% fill factor" (no wasted space) and super-precise location data.

2. The Challenge: How to Test Without a Giant Collider

To test these sensors, scientists usually need a massive particle accelerator (a test beam) that shoots protons at them. But these machines are huge, expensive, and hard to book. You can't just run a quick test whenever you want.

The researchers asked: "Can we use a laser instead?"
A laser is like a tiny, precise flashlight. If you shine it on the sensor, it creates a spark of electricity, just like a particle would. It's cheap, fast, and you can move it around with a robot arm to test every single inch of the sensor.

The Catch: A laser creates a spark differently than a real particle. A real particle (a proton) is like a heavy bowling ball dropping into the sensor, creating a chaotic, unpredictable splash of energy. A laser is like a precise water dropper. The researchers needed to prove that even though the method of making the spark is different, the result (how well the sensor measures time and position) is the same.

3. The Experiment: The "Calibration" Dance

The team built a setup with a laser, a robot arm, and high-speed cameras (oscilloscopes) to watch the sensor react. They compared their laser results against data they already had from a real 120 GeV proton beam.

They found a few hurdles:

• The Noise Floor: The laser setup was a bit "noisier" (like trying to hear a whisper in a windy room) compared to the proton beam setup.
• The Calibration: They had to tune the laser's brightness so that the "splash" it made matched the "splash" of a real proton.

Once they adjusted for the noise and matched the signal strength, they compared the two methods.

4. The Results: Lasers and Protons Agree!

The big news is that the laser and the proton beam gave the same answers.

• Position: Both methods could pinpoint the hit location with the same incredible accuracy (about 10 micrometers, which is thinner than a human hair).
• Time: Both methods measured the time of the hit with similar precision (around 30–40 picoseconds, which is 0.00000000003 seconds).

This is huge because it means scientists can now use these laser setups in their own labs to test new sensors quickly, without waiting for a turn at a giant particle accelerator. It speeds up research and development significantly.

5. The Mystery: The "Ghost" in the Machine

While the laser and proton results matched on position and timing, the researchers noticed a small mystery.

• The Jitter: In physics, "jitter" is like the slight wobble in a stopwatch. They calculated how much wobble should exist based on the signal strength.
• The Extra Wobble: Even after accounting for the known wobble, there was still a tiny bit of extra "wobble" (time uncertainty) in the laser data that they couldn't fully explain. It's like timing a race and knowing the runner's speed and the wind, but the clock still seems to be off by a fraction of a second for reasons you can't see.

They used computer simulations (digital twins of the sensor) to try to understand this, but the mystery remains. This suggests there are other, subtle factors affecting how these sensors measure time that we don't fully understand yet.

The Bottom Line

This paper is a success story for efficiency.

• Before: To test a new sensor design, you had to wait for a slot at a massive, expensive particle accelerator.
• Now: You can build a laser setup in your lab, calibrate it against one known proton test, and then trust it to test hundreds of new designs quickly.

It's like realizing that while a wind tunnel is great for testing cars, you can actually get 95% of the same data using a powerful fan in your garage, as long as you know how to calibrate the fan. This allows scientists to innovate faster, bringing us closer to the next generation of particle detectors that can see the universe with incredible speed and clarity.

Technical Summary

1. Problem Statement

As high-energy physics experiments (such as CMS and ATLAS at the HL-LHC) move toward higher luminosities, there is an urgent need for 4D tracking detectors that provide both precise spatial and temporal resolution (on the order of 10 ps) to separate out-of-time pileup tracks.

• Technology: AC-coupled Low Gain Avalanche Diodes (AC-LGADs) are a promising candidate for 4D tracking due to their internal gain, thinness, and 100% fill factor (achieved by replacing segmented n+ layers with a continuous resistive layer).
• Challenge: While testbeam facilities using Minimum Ionizing Particles (MIPs) are the standard for characterizing these sensors, they are resource-intensive and limited in availability. There is a need for an efficient, in-house alternative to accelerate R&D.
• Specific Issue: Laser sources deposit charge differently than MIPs (photons vs. ionization tracks), and experimental setups often suffer from noise discrepancies. It is unclear if laser-based measurements can accurately replicate MIP-based performance metrics (spatial and time resolution) after proper calibration.

2. Methodology

The authors developed a comprehensive experimental and simulation framework to compare laser-based measurements with 120 GeV proton testbeam data.

Experimental Setup

• Device Under Test (DUT): Strip-type AC-LGAD sensors (1 cm long, 50 µm wide strips, 500 µm pitch) fabricated by Hamamatsu Photonics (HPK). Three specific sensors (S1, S2, S4) with varying active thicknesses (20 µm and 50 µm) and sheet resistances were tested.
• Laser Source: A pulsed infrared laser (1060 nm) with low timing jitter (< 3 ps) was used to generate signals emulating MIP responses.
• Readout: Signals were read via a 16-channel Fermilab board with two-stage amplifiers and captured by high-bandwidth oscilloscopes (Keysight MSO7104B and Teledyne LeCroy Waverunner 8208HD).
• Scanning: A motorized XY Z stage performed 2D scans with high precision (0.1 µm) to map sensor response across the active area.

Calibration & Analysis

• Bias Voltage: Determined by monitoring noise levels to find the breakdown point (operating 1 V below breakdown).
• Intensity Tuning: Laser intensity was adjusted so that the mid-gap signal amplitude (signal shared equally between two strips) matched the Most Probable Value (MPV) of MIP-induced signals.
• Reconstruction Algorithms:
  • Position: Interpolated using the amplitude fraction ($f = a_1 / (a_1 + a_2)$) of the two leading strips.
  • Time: A "multi-channel timestamp" was calculated using a weighted average of the Time of Arrival (ToA) from leading and sub-leading channels, weighted by the square of their amplitudes ($t = \frac{a_1^2 t_1 + a_2^2 t_2}{a_1^2 + a_2^2}$).
• Noise Correction: A critical step involved normalizing laser results to MIP results based on the ratio of baseline noise levels ($N_{MIP} / N_{Laser}$) to account for setup-induced differences (cabling, grounding, temperature).

Simulation Studies

• Tools: Used Silvaco TCAD for 2D device/electric field simulation and Weightfield2 (WF2) for waveform and signal propagation simulation.
• Purpose: To validate the theoretical jitter formula ($\sigma_{jitter} = N / (dV/dt)$) and understand the contributions of noise, sampling rate, and charge deposition to total time resolution.

3. Key Contributions

1. Validation of Laser Calibration: Demonstrated a robust methodology to calibrate an infrared laser source to mimic MIP interactions, specifically by matching mid-gap amplitudes and correcting for noise differences.
2. Noise Normalization Framework: Identified that baseline noise differences between laser and testbeam setups were the primary cause of discrepancies in resolution. The paper provides a normalization factor to reconcile these datasets.
3. Simulation-Driven Jitter Analysis: Developed a simulation framework to compare predicted jitter against observed time resolution, revealing that standard jitter formulas may require scale factors depending on sampling rates and CFD thresholds.
4. Identification of "Additional Components": Isolated a non-negligible residual in time resolution that cannot be explained by jitter or Landau fluctuations alone, suggesting unknown systematic effects in laser-based measurements.

4. Key Results

• Spatial Resolution: After normalizing for noise, the position resolution obtained via the laser source was compatible with 120 GeV proton beam results across all three sensors (S1, S2, S4). The resolution followed the expected theoretical curve based on signal sharing and noise.
• Temporal Resolution (Weighted Jitter):
  • The weighted jitter (the dominant term in time resolution) showed good agreement between laser and MIP sources after noise normalization.
  • Risetime: The rise times of the signals were comparable between the two sources once amplitudes were matched.
• Discrepancies:
  • Despite matching jitter, a non-negligible residual remained in the total time resolution for both laser and MIP data (especially in the 20 µm sensor, S4).
  • Simulation studies indicated that the standard single-channel jitter formula overestimates the observed resolution, requiring a "scale factor" to align theory with experiment.
  • The residual suggests the presence of an "additional component" (non-jitter, non-Landau) that affects timing, the origin of which remains under investigation.

5. Significance

• Accelerated R&D: This work validates that in-house laser setups can reliably replace or augment expensive and time-consuming testbeam campaigns for AC-LGAD characterization. This allows for rapid iteration of sensor designs (e.g., varying sheet resistance or thickness).
• 4D Tracking Viability: Confirms that AC-LGADs maintain high spatial and temporal performance, reinforcing their status as a leading candidate for future collider upgrades (e.g., Electron-Ion Collider).
• Methodological Refinement: The paper highlights the critical importance of noise characterization and calibration when comparing different particle sources. It also points to the need for refined theoretical models of time resolution that account for factors beyond simple jitter (such as signal propagation delays and sampling effects).

In conclusion, the paper successfully bridges the gap between laser-based characterization and high-energy testbeam data, providing a reliable, scalable tool for the development of next-generation 4D tracking sensors.