🔭 astrophysics

Output-Stage Design Optimization for High-Sensitivity SiSeRO CCDs and SiSeRO Active Pixel Sensors

This paper presents device simulations and design optimizations for next-generation SiSeRO output stages to enhance noise and speed performance, while outlining the architectural updates required to develop SiSeRO active pixel sensors that combine the superior X-ray performance of CCDs with the benefits of active pixel architectures.

Original authors: Tanmoy Chattopadhyay, Sven Herrmann, Kevan Donlon, Ilya Prigozhin, Peter Orel, Steven W. Allen, Marshall W. Bautz, Michael Cooper, Catherine E. Grant, Beverly LaMarr, Christopher Leitz, Eric D. Miller

Published 2026-07-28

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Tanmoy Chattopadhyay, Sven Herrmann, Kevan Donlon, Ilya Prigozhin, Peter Orel, Steven W. Allen, Marshall W. Bautz, Michael Cooper, Catherine E. Grant, Beverly LaMarr, Christopher Leitz, Eric D. Miller, R. Glenn Morris, Abigail Y. Pan, Tonya L. Peshel, Artem Poliszczuk, Gregory Prigozhin, Haley R. Stueber, Keith Warner

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the universe whispering its secrets in X-rays, a high-energy language that reveals black holes, exploding stars, and the invisible scaffolding of galaxies. To hear these whispers clearly, astronomers need eyes that are not just big, but incredibly sensitive and fast. They need detectors that can count individual particles of light with the precision of a master accountant, yet do it thousands of times a second without getting tired or making mistakes. This is the world of X-ray astronomy, where the goal is to build "megapixel" cameras that can see the faintest glimmers of energy without adding their own static to the signal.

The challenge is a bit like trying to hear a pin drop in a hurricane. Traditional cameras are either fast but noisy, or quiet but slow. A new technology called SiSeRO (Single electron Sensitive Read Out) was invented to solve this, acting like a super-sensitive microphone that can detect the tiny electrical "ping" of a single electron. However, even this amazing microphone has a bit of static in its wiring, and the signal it produces could be louder. This paper is about tuning that microphone's internal wiring to make the signal crystal clear and the noise vanish, so we can build the ultimate cosmic camera for future space telescopes.

The Cosmic Microphone Tuning Session

Think of the SiSeRO detector as a high-tech microphone designed to listen to the universe. Inside this microphone, there is a tiny "listening post" (the internal gate) where signal electrons gather. Sitting right above this post is a transistor, which acts like a faucet. When electrons pile up in the listening post, they change the pressure, and the faucet opens or closes to let water (current) flow. The more electrons you have, the more water flows. The goal of this research was to make that faucet incredibly sensitive to even the tiniest drop of water, while ensuring the pipes don't rattle with their own noise.

The researchers used a powerful computer program called Sentaurus TCAD to build a virtual version of this faucet and tweak its design. They weren't just guessing; they were running simulations to see how changing the shape and size of the parts would affect the performance.

Making the Faucet More Sensitive
The first thing they looked at was how to make the transistor more sensitive to the electrons. They found two main levers to pull:

The Buried Channel: Imagine the path the water takes as a road. By adding more "traffic" (doping) to this road, they made it wider and smoother, allowing the water to flow more easily. This boosted the sensitivity, or "transconductance," of the device.
The Gate Oxide: This is a thin layer of insulation between the control knob and the road. The team found that making this layer thinner was like tightening the connection between the knob and the road. A thinner layer meant the knob had more control, significantly boosting the sensitivity.

By combining these tweaks, they simulated a design that could achieve a transconductance (sensitivity) of over 16 µS, a big jump from the original 9 µS.

Turning the Signal into a Loud, Clear Voice
Sensitivity is great, but you also need a strong signal. The team wanted to know: if one electron drops into the listening post, how much does the water flow increase? This is called "gain." In their original design, one electron caused the current to jump by 511 pA (picoamperes).

Through their simulations, they discovered that the shape and location of the "trough" (the hole where electrons sit) mattered a lot.

Where to put the hole: They found that moving the electron-holding hole closer to the "drain" (the exit of the water) made the biggest difference. It's like placing a magnet right at the exit of a pipe; it pulls the water through much harder. When the electrons were near the drain, the gain was highest.
How big the hole should be: If the hole was too small, it didn't catch enough of the signal. If it was too big, it created too much "capacitance" (like a heavy bucket that's hard to lift), which slowed things down. They found a "Goldilocks" size that maximized the signal.
The result: With these optimizations, the simulated gain jumped from 511 pA/e to roughly 1100 pA/e. That means the same single electron now produces more than double the signal!

Silencing the Static
Even with a loud signal, you don't want the background hiss of the pipes. The team analyzed the "noise" in the system. They found that the main culprit for low-frequency noise (the deep, rumbling static) was "bulk defects"—tiny imperfections inside the silicon material itself. It's like having a few bad spots in the pipe that cause turbulence.

They also looked at the "polysilicon gate," which is like the metal casing of the faucet. They discovered that if this casing was made of a material that conducted electricity better (heavily doped), the thermal noise (the heat-induced rattling) dropped significantly. By switching to a more conductive gate, they simulated a noise level of about 0.5 pA/√Hz, which is several times quieter than the older, less conductive designs.

The Future: A Stereo System for Space
The paper doesn't just stop at making a better single microphone; it looks ahead to building a whole stereo system. The researchers are designing a new type of sensor called an "Active Pixel Sensor" (APS) that uses two of these SiSeRO microphones side-by-side in every single pixel.

Imagine having two microphones for every spot in your camera. You can move the signal from one to the other, measure it, move it back, and measure it again. This technique, called Repetitive Non-Destructive Readout (RNDR), allows the camera to average out the noise, potentially getting down to less than half an electron of noise. The simulations suggest that to make this work, the "troughs" holding the electrons need to be deeper and more confined so the charge doesn't spill over when it's being shuffled between the two microphones.

The Bottom Line
This paper is a roadmap for the next generation of cosmic cameras. By simulating changes in the transistor's geometry and materials, the team has shown that it is possible to double the signal strength and significantly cut the noise. While these results are currently based on computer simulations and not yet built in a lab, they provide a clear path forward. If these designs are built, future X-ray telescopes could capture images of the universe with unprecedented clarity, counting every single electron with a speed and silence that was previously impossible.

Technical Summary: Output-Stage Design Optimization for High-Sensitivity SiSeRO CCDs and SiSeRO Active Pixel Sensors

Problem Statement
Future X-ray astronomy missions require detectors capable of simultaneously achieving high frame rates, low read noise, and large imaging formats. While advanced X-ray charge-coupled devices (CCDs) offer superior noise and energy resolution compared to conventional CCDs, their frame rates remain limited by serial readout architectures. Conversely, CMOS-based Active Pixel Sensors (APS) offer high-speed operation but typically exhibit higher noise levels or lack sufficient energy bandwidth. The Single electron Sensitive Read Out (SiSeRO) technology, developed at MIT Lincoln Laboratory, addresses these limitations by enabling sub-electron noise performance through Repetitive Non-Destructive Readout (RNDR). However, the performance of current SiSeRO devices is constrained by output-stage gain and excess low-frequency noise. Optimizing these characteristics is critical for developing next-generation high-sensitivity SiSeRO CCDs and for extending the technology to large-format APS suitable for flagship missions such as Lynx.

Methodology
The authors utilized two-dimensional Sentaurus TCAD simulations to investigate the dependence of SiSeRO output-stage performance on key process and device parameters. The study focused on the p-type buried-channel MOSFET output transistor fabricated above an internal gate region.

Simulation Framework: A 2D model was constructed using Synopsys Sentaurus, replicating the fabrication process of first-generation SiSeRO CCDs. The geometry included a p-substrate, p-type buried MOSFET channel, polysilicon gate, and an internal gate trough structure.
Electrical Simulations: Current-voltage characteristics were simulated using thermodynamic transport models at 300 K, incorporating field-dependent mobility, quantum corrections, and Shockley-Read-Hall recombination. Transconductance ( $g_m$ ) was extracted from transfer characteristics, and detector gain ( $G_q$ ) was evaluated by introducing discrete charge into the internal gate and calculating the resulting change in drain current.
Noise Simulations: Noise analysis was performed using the Impedance Field Method (IFM), accounting for thermal diffusion noise, interface trap noise, and bulk defect noise across a frequency range of 1 kHz to 10 MHz.
Parameters Studied: The optimization study varied MOSFET channel implant dose, gate oxide thickness, trough implant energy, trough size, trough location, interface trap density, bulk defect density, and polysilicon gate doping.

Key Contributions and Results
The paper presents a systematic optimization of the SiSeRO output stage, identifying specific design pathways to enhance gain and reduce noise.

Transconductance Optimization:
- Buried-Channel Implant Dose: Increasing the dose improves channel conductivity and current density, leading to higher transconductance, though gains plateau at higher doses due to increased scattering.
- Gate Oxide Thickness: Thinner gate oxides significantly improve electrostatic coupling between the gate and the buried channel.
- Result: The optimized design achieves transconductance values exceeding 16 µS, a substantial improvement over the original structure's ~9 µS.
Gain Optimization:
- Channel Implant Dose: Gain increases monotonically with channel dose, though operational limits exist to prevent excessive current density.
- Trough Implant Energy: Lower implant energies improve gain by increasing the change in hole velocity in the presence of stored electrons.
- Trough Size: Gain exhibits a non-monotonic dependence on trough size; an intermediate width yields maximum gain by balancing capacitance and charge coupling area.
- Trough Location: Maximum gain is achieved when the internal gate trough is positioned closest to the drain side of the channel. This placement maximizes the electric field enhancement and hole acceleration toward the drain.
- Result: The optimized configuration increases detector gain from approximately 511 pA/e (baseline) to roughly 1100 pA/e.
Noise Analysis:
- Low-Frequency Noise: Bulk defect states were identified as the dominant source of low-frequency ( $1/f$ ) noise. Reducing these defects is critical for future fabrications.
- Thermal Noise: Thermal noise is strongly influenced by polysilicon gate resistance. Simulations demonstrate that using heavily doped polysilicon gates significantly reduces thermal noise compared to undoped gates.
- Result: The optimized structure achieves a simulated thermal-noise spectral density of approximately 0.5 pA/ $\sqrt{\text{Hz}}$ , a several-fold improvement over undoped designs.

Significance and Future Work
The paper claims that these simulation-based optimizations provide a clear pathway to enhance the sensitivity and speed of SiSeRO devices. The primary significance lies in the ability to double the detector gain while simultaneously improving transconductance and reducing noise, which are essential for meeting the stringent requirements of future X-ray missions.

Furthermore, the authors discuss the implications for SiSeRO Active Pixel Sensors (APS). They propose a dual-SiSeRO pixel architecture where two sensing nodes are connected via a common electron-transfer channel, enabling RNDR operation through simultaneous measurement of signal and reference states. This design aims to combine the proven X-ray performance of CCDs with the architectural advantages of APS, such as region-of-interest readout and improved radiation hardness.

Future work outlined in the paper includes:

Developing three-dimensional TCAD process simulations for dual-SiSeRO pixels.
Investigating deeper trough potentials and increased implant doses to ensure charge confinement during transfer operations in APS devices.
Validating the dual-SiSeRO RNDR operation and optimizing channel width and sense node isolation.

The authors conclude that these simulations form the foundational basis for the design and development of future large-format, single-electron-sensitive, fast imaging detectors.

The Cosmic Microphone Tuning Session

More like this