Systematic Investigation of Acceptor Removal in HPK LGADs with Modified Gain Layers

This study investigates various modified gain-layer designs in HPK LGAD prototypes to combat radiation-induced acceptor removal, finding that carbon implantation is the only effective method for improving radiation tolerance among the approaches tested.

The Gist

The Story of the "Fading Battery": Making High-Tech Sensors Last Longer

Imagine you are building a super-fast, high-precision stopwatch for a futuristic race. This stopwatch isn't just a piece of plastic; it’s a tiny silicon sensor called an LGAD.

To work perfectly, this sensor needs a "boost" of electricity—kind of like a built-in turbocharger—to make sure it can detect particles moving at nearly the speed of light. This turbocharger is powered by a special layer inside the sensor called the "Gain Layer."

The Problem: The "Fading Battery" Effect

Here is the catch: these sensors are going to be used in massive particle colliders (like the ones at CERN). These machines are incredibly violent environments, constantly bombarded by radiation.

Think of radiation like tiny, invisible sandstorms. As the sandstorms hit the sensor, they slowly clog up the turbocharger. In scientific terms, this is called "Acceptor Removal."

Imagine the turbocharger is a battery that provides the extra "oomph" the sensor needs. The radiation acts like a slow leak, draining the battery's capacity. Eventually, the battery gets so weak that the sensor can't "click" fast enough to keep time. If the sensor can't keep time, the whole experiment fails.

The Experiment: Finding the Best "Shield"

Scientists wanted to find a way to stop this "battery drain." They worked with a company called Hamamatsu to create several different "shielding" designs to see which one could withstand the sandstorm best. They tried three main strategies:

1. The Oxygen Clean-up (The "Dusting" Method): They thought maybe oxygen inside the sensor was making the damage worse, so they tried to make the sensor "cleaner" by reducing oxygen.
2. The Carbon Armor (The "Buffer" Method): They tried adding Carbon to the mix. Think of this like adding a layer of sacrificial padding. If the radiation hits the carbon instead of the "battery," the battery stays strong.
3. The Compensation Trick (The "Double-Battery" Method): They tried adding a second type of chemical (Phosphorus) to balance things out, hoping that as one chemical faded, the other would step in to help.

The Results: What Actually Worked?

After putting these sensors through intense radiation tests (using protons and neutrons), the scientists found a clear winner:

• The Oxygen Clean-up? It didn't really help. It was like trying to clean a room while a sandstorm is still blowing in; the oxygen wasn't the real problem.
• The Compensation Trick? It was a bit of a mess. It didn't work the way they expected. It turns out that the two chemicals they added started interfering with each other, like trying to balance a scale where the weights keep changing shape.
• The Carbon Armor? This was the superstar. The carbon worked exactly like the "sacrificial padding" theory suggested. The radiation hit the carbon, and the carbon "took the hit," leaving the main power source (the boron) much more intact. The sensors with carbon stayed much more powerful for much longer.

Why Does This Matter?

In the future, we are going to build even bigger, more powerful machines to study the universe. To do that, we need sensors that don't "die" the moment things get intense.

This paper tells engineers: "If you want your high-speed sensors to survive the cosmic sandstorm, don't worry so much about oxygen or complex balancing acts—just add some Carbon armor." It’s a roadmap for building the eyes and ears of the next generation of physics discovery.

Technical Summary

Technical Summary: Systematic Investigation of Acceptor Removal in HPK LGADs with Modified Gain Layers

1. Problem Statement

Low-Gain Avalanche Diodes (LGADs) are critical for the 4D tracking requirements of future high-energy hadron colliders (like the HL-LHC and FCC-hh), providing high spatial resolution and exceptional timing precision ($\mathcal{O}(10)$ ps). However, their performance is severely limited by acceptor removal in the $p^+$ gain layer. Under heavy radiation, electrically active boron acceptors are deactivated through defect-mediated reactions, reducing the effective acceptor concentration ($N_A$). This leads to a decrease in the internal electric field and a loss of signal gain, necessitating higher operation voltages to recover timing performance—a strategy limited by destructive breakdown phenomena like single-event burnout (SEB).

2. Methodology

The study investigated several design strategies to mitigate this degradation using prototype sensors produced by Hamamatsu Photonics K.K. (HPK). The researchers tested three primary modification approaches:

• Oxygen Modification: Reducing oxygen contamination (to limit $B_iO_i$ defect formation) and testing "Partially Activated Boron" (PAB) as a sacrificial component to clean oxygen-related reactions.
• Carbon Implantation: Co-doping carbon into the gain layer to act as a sink for silicon interstitials, theoretically preventing them from deactivating boron.
• Gain-Layer Compensation: Co-implanting donors (phosphorus) with acceptors (boron) to moderate the reduction of the effective net doping ($N_{eff} = N_A - N_D$) under irradiation.

Experimental Setup:

• Irradiation: Samples were subjected to proton irradiation (45 MeV and 70 MeV at RARiS) and reactor-neutron irradiation (JSI, Slovenia).
• Characterization:
  • Electrical: $I$-$V$ and $C$-$V$ measurements were used to extract the gain-layer depletion voltage ($V_{gl}$) and the acceptor-removal coefficient ($C_A$).
  • Timing: $^{90}\text{Sr}$ beta-ray measurements were used to determine timing resolution ($\sigma_t$) and the operation voltage ($V_{op}$) required to achieve optimal timing.

3. Key Results

• Carbon Implantation is Superior: This was the only approach that provided a significant improvement in radiation tolerance. Carbon-implanted samples ($B+C$ and $B+C+O$) showed substantially lower $C_A$ values and required significantly lower $V_{op}$ to maintain timing performance compared to standard boron-only ($B$) samples.
• Oxygen Modification is Ineffective: Neither the reduction of oxygen nor the PAB approach yielded significant improvements. This suggests that $B_iO_i$ complexes are not the dominant mechanism for acceptor removal in these specific HPK structures.
• Compensation Fails the "Independent-Removal" Model: The expectation that donor removal would offset acceptor removal did not hold. Compensation-only samples behaved similarly to standard boron-only samples. Furthermore, combining compensation with carbon ($2B+1P+C$) offered no clear advantage over carbon-only implantation, suggesting that acceptor and donor removal processes are coupled through shared defect kinetics (e.g., vacancy/interstitial competition).
• Particle/Energy Dependence: The acceptor-removal coefficient $C_A$ is not a universal constant; it depends on the irradiation particle type and energy. $C_A$ values for reactor neutrons were systematically lower than those for 70 MeV protons.

4. Key Contributions and Significance

• Validation of the Interstitial Sink Model: The success of carbon implantation provides strong experimental evidence for the theory that silicon interstitials (the drivers of boron deactivation) can be captured by carbon, thereby preserving the active boron concentration.
• Refinement of Design Strategies: The paper provides a "negative" result for oxygen reduction and compensation, saving future R&D resources by steering designers away from these ineffective paths for this specific doping regime.
• Practical Design Guidelines: The authors conclude that while carbon implantation is the most promising path, it must be optimized to balance radiation hardness against the increased leakage current and earlier breakdown observed in non-irradiated carbon-implanted samples.
• Impact on Detector Modeling: By demonstrating the dependence of $C_A$ on particle energy and type, the study highlights the necessity of using realistic particle spectra (rather than a single universal $C_A$) when simulating the lifetime and performance of future collider detectors.