🔬 physics

Development of a Neutron Imager using CMOS Pixelated Sensor

This paper reports the successful imaging of a Siemens Star Chart using a 10B-INTPIX4 CMOS pixelated sensor at J-PARC, characterizing the system's performance with a 2.7 mrad beam divergence (equivalent to an L/D ratio of 370) and a 50% modulation transfer function cutoff at 12 line pairs per millimeter.

Original authors: Yoshio Kamiya, Hidetoshi Ohshita, Ryutaro Nishimura, Toshiya Muto, Shingo Mitsui, Tomohiro Seya, Izumi Umegaki, Yasuo Arai

Published 2026-06-17

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Yoshio Kamiya, Hidetoshi Ohshita, Ryutaro Nishimura, Toshiya Muto, Shingo Mitsui, Tomohiro Seya, Izumi Umegaki, Yasuo Arai

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 trying to take a sharp photograph of a tiny, intricate gear using a flashlight that isn't perfectly steady. That's essentially what this team of scientists did, but instead of a camera and a flashlight, they used a high-tech sensor and a beam of neutrons (tiny, invisible particles) to take a picture of a special test pattern called a "Siemens Star."

Here is the story of how they did it and what they found, broken down into simple concepts.

The Goal: A Super-Sharp "Neutron Camera"

Neutrons are amazing for looking inside materials (like seeing the bones of a machine without taking it apart), but they are hard to catch. To take a picture with them, you need a sensor that can turn these invisible particles into a visible image.

The researchers built a new kind of "neutron camera" using a CMOS sensor—the same type of chip found in your smartphone camera, but super-charged. To make it sensitive to neutrons, they coated the back of the sensor with a very thin layer of Boron-10. Think of this layer like a sticky trap: when a neutron hits it, it explodes into two charged particles (like a tiny alpha particle and a lithium atom) that the sensor can easily see and record.

The Test: The "Siemens Star"

To see how good their new camera was, they didn't just take a picture of a random object. They used a Siemens Star Chart.

The Analogy: Imagine a pizza cut into 128 tiny slices, but instead of being solid, the slices are alternating black and white lines that get closer and closer together as you move toward the center.
The Challenge: If your camera is blurry, the lines near the center (where they are very close together) will just look like a gray smear. If your camera is sharp, you can still see the individual lines.

They placed this chart in front of their sensor at a massive research facility called J-PARC in Japan, which shoots out a powerful beam of neutrons.

The Process: Sorting the Noise

Neutrons aren't the only things flying around; there are also background "noise" particles (like gamma rays).

The Filter: The scientists wrote a computer program to act like a bouncer at a club. It looked at the "footprints" left by particles on the sensor.
- Neutrons leave a neat, round, symmetrical footprint (like a small puddle).
- Background noise leaves a long, stretched-out footprint (like a smear).
The computer kept only the neat, round footprints and threw away the smears, ensuring the final picture was made only of true neutron data.

The Results: How Sharp Was the Picture?

When they developed the image, they found some interesting things:

The Blur: The image wasn't perfectly sharp; it was slightly fuzzy. They measured this fuzziness and found it was about 17 micrometers wide (that's about the width of a human hair).
The Cause: They realized this blur wasn't because their camera was bad. Their camera is actually very sharp (better than 5 micrometers). The blur was actually caused by the neutron beam itself.
- The Analogy: Imagine shining a flashlight through a small hole. If the hole is perfect, the beam is a straight, tight laser. But if the hole is slightly wobbly or the light source is wide, the beam spreads out a little. The neutrons were spreading out slightly as they traveled, causing the image to blur.
The "F-Stop" of Neutrons: In photography, the "f-number" tells you how much light is coming in. In neutron imaging, they use something called the L/D ratio (Length to Diameter). Their experiment showed that the beam was behaving as if it had an L/D ratio of 370. This is a very high number, meaning the beam was quite tight and well-controlled, which is great for getting clear pictures.

The "Cutoff" Point

They also tested how small the details could be before they disappeared.

They looked at the Siemens Star and asked: "At what point do the lines become so close together that we can no longer tell them apart?"
They found that the system could clearly distinguish 12 pairs of lines per millimeter before the image dropped to 50% clarity. This is a measure of the camera's "resolution" or sharpness.

The Bottom Line

The scientists successfully built a new, high-precision neutron camera using a standard computer chip modified with a boron coating. They proved it works by taking a picture of a test star pattern.

The picture was slightly blurry, but not because the camera was weak. The blur came from the neutron beam itself spreading out slightly. By measuring this blur, they confirmed their camera is incredibly sharp (capable of seeing details smaller than a human hair) and that the neutron beam they used was very high quality. This tool could help scientists monitor the size and shape of neutron beams in places that are too dangerous for humans to enter, like right next to the neutron source.

Technical Summary: Development of a Neutron Imager Using CMOS Pixelated Sensor

Problem and Objective
With the advancement of high-intensity neutron sources, there is a growing need for high-precision neutron imaging tools to support materials research and industrial applications. While synchrotron radiation facilities have driven industrial progress, similar advancements in neutron facilities are required to expand their societal utility. The authors aim to develop a high-precision neutron imager based on a CMOS pixel sensor capable of achieving sub-5 µm resolution when combined with thin conversion materials. This paper presents the experimental evaluation of such a system using a standard imaging performance tool, the Siemens Star Chart, to quantify spatial resolution and modulation transfer function (MTF).

Methodology
The study utilized the 10B-INTPIX4, a monolithic CMOS pixelated imager based on the SOIPIX-INTPIX4 sensor family. Key technical specifications include:

Sensor Architecture: Fabricated using a 0.2 µm CMOS FD-SOI process, featuring a 17×17 µm² pixel size, an active area of 14.1×8.7 mm² (832×512 pixels), and a 300 µm wafer thickness. The design integrates front-end circuits on a semiconductor layer above an insulator, eliminating bump bonding and reducing parasitic capacitance.
Neutron Conversion: A 200 nm thick enriched boron ( $^{10}$ B) layer was deposited on the sensor's backside via Ar sputtering. This layer converts neutrons into charged particles ( $\alpha$ and $^7$ Li) through $^{10}$ B(n, $\alpha$ ) $^7$ Li interactions.
Data Acquisition: Signals were read out using the PF-DAQSIX system developed at KEK, featuring Kintex-7 FPGAs, 12-bit ADCs, and a 10 Gbps Ethernet interface.
Experimental Setup: The experiment was conducted at J-PARC MLF Beam Line 21 (NOVA). The imager was positioned 21 m from the neutron production target (6.1 m downstream from the standard sample location). A 20 mm collimator 6.5 m upstream established a geometrical L/D ratio of 330.
Target: A Siemens Star Chart, consisting of a 5-µm-thick Ge film with etched regions arranged in 128 radial segments and six concentric donut-shaped regions, was placed 6 mm upstream of the sensor.
Event Selection: Neutron events were distinguished from background $\gamma$ -rays by analyzing cluster shapes within a 7×7 pixel region. Events were filtered based on total charge ( $m_0$ ) and the second moment of the charge distribution ( $m_2$ ), selecting a range of 1400 < $m_0$ < 7000 channels to isolate single neutron interactions.

Key Results

Spatial Resolution: The image blurriness was characterized by the root mean square (rms) of the line spread function (LSF). The analysis yielded a resolution of 1.0 pixel, equivalent to 17 µm.
Beam Divergence and Geometry: Assuming the observed blurriness is primarily due to beam divergence, the rms beam divergence was calculated to be 2.7 mrad. This implies an effective neutron beam size of approximately 18 mm at the collimator, resulting in an effective geometrical ratio (L/D) of 370, which aligns closely with the NOVA design specification of 330.
Contrast and MTF: The Modulation Transfer Function (MTF) was evaluated by analyzing image contrast as a function of structural scale. The maximum contrast ( $\eta_0$ ) was found to be 0.29, corresponding to a 45% attenuation in the 5 µm Gd foil. The structure corresponding to the 50% MTF cutoff was determined to be 12 line-space pairs per mm.

Significance and Claims
The paper demonstrates that the 10B-INTPIX4 imager successfully captures high-resolution images of a Siemens Star Chart, validating its capability for precise neutron imaging. The authors claim that the system's performance is consistent with theoretical expectations for beam divergence and sensor resolution.

A primary significance highlighted is the potential for this device to be used in environments inaccessible to humans due to high radiation levels. By configuring the system to capture the entire target plane, it can effectively estimate or monitor neutron beam sizes at locations such as neutron production targets. The results confirm that the sensor, when paired with a thin conversion layer, is a viable candidate for achieving high-resolution neutron detection, supporting the broader goal of enhancing user environments at neutron facilities.