A digital camera sensor, such as a CMOS (Complementary Metal-Oxide Semiconductor), is an electronic device designed to convert photons of light into electrical signals. However, these sensors are not only sensitive to photons. They can also respond to other excitation sources, including charged particles from cosmic rays. As part of a research project investigating the interaction of cosmic rays with CMOS sensors, we collected over 2,000 hours of dark images to build a sample of traces left by passing particles. Below, we explore the basics of the vast world of CMOS particle detection.

Sensor principle and interaction with charged particles

The conversion of photons into electrical signals relies on photodiodes, present in each pixel of the sensor matrix. Their operation is based on the photoelectric effect: the photodiode, made of silicon, is a light-sensitive semiconductor. When photons strike the photodiode, their energy excites the silicon atoms, allowing electrons to leave their initial energy levels and escape from their atomic layers. These freed electrons are collected by a transistor, which converts the accumulated charge into an electrical voltage signal. The amount of charge collected is proportional to the number of photons that hit the photodiode, and thus to the light intensity.

The use of the photoelectric effect in photodiodes also makes them sensitive to charged particles, enabling the detection of phenomena such as the passage of cosmic rays. Since these particles carry an electric charge, they interact directly—through electromagnetism—with the silicon of the photodiodes as they pass through the sensor. By crossing the silicon substrate, the particle disturbs the silicon atoms, tearing electrons from their electron clouds. Cosmic rays, which often have very high kinetic energy, can release hundreds or even thousands of electrons as they pass through a photodiode. They therefore leave bright traces on the sensor, often in the form of lines that mark their trajectories.

Origin of cosmic rays

Cosmic rays are charged particles coming from space that continuously bombard the Earth. These rays, made up mainly of protons (about 90%), travel at relativistic speeds and carry enormous kinetic energy. Their exact origin remains uncertain, but several phenomena have been identified as direct sources: the Sun, objects emitting astrophysical jets (blazars, pulsars), and supernova shock waves.

When a cosmic ray enter the upper layers of Earth’s atmosphere, they collide with the nuclei of atoms present in the air, mainly nitrogen and oxygen. These high-energy collisions produce a cascade of secondary particles in a process called atmospheric shower (or air shower): when a proton (or another nucleus) strikes a nitrogen or oxygen nucleus, it transfers part of its energy to it, breaking it apart into subatomic particles such as pions (π) and kaons (K). Pions and kaons, which are unstable particles, decay almost immediately. Neutral pions (π0) decay into photons (producing gamma radiation), while charged pions (π+ and π−) decay into muons and neutrinos. Kaons can produce pions (π0, π+, and π−) as well as muon-neutrino pairs.

The particles produced in these showers (muons, photons, neutrinos, and electron-positron pairs) are the ones that can be observed with CMOS sensors at sea level (except for neutrinos. Being neutral particles, they don't interact through electromagnetism and thus do not trigger the pixels of the CMOS). Muons, in particular, are massive particles. At their creation, they have very high speeds and thus high energies, which gives them great inertia and the ability to penetrate deeply into matter. They are therefore the main component of cosmic radiation.

Observation

A cosmic ray observation session using a CMOS sensor simply involves taking dark frames with a long exposure time, preferably with a cooled camera. In the resulting images, two main types of tracks are generally observed: straight tracks, characteristic of cosmic muons, and irregular tracks, associated with lighter particles such as electrons from the surrounding radiation.

The length of the track left by a muon depends entirely on the angle at which it crosses the sensor. Thus, some muons arriving at a very oblique angle can produce exceptionally long tracks, while those arriving perpendicularly produce tracks that appear as spots:

Electrons coming from local radiation sources (for example, beta radiation) often leave irregular tracks. Their low mass and moderate speed make them particularly sensitive to magnetic fields present around the sensor. As a result, irregular worm-like patterns can be observed:

What's the project about

While running this type of experiment, you'll find other types of tracks, resulting from rarer interactions (for example, pair production). These interactions and signatures are poorly modeled, but knowing precisly which particle type (muon, electron, photon, proton, pion, kaon...) produced the track is essential to calculate the energy from the track's brightness. The project's goal is to reliably identify the particle type based on track morphology. The integration of thousands of hours of dark calibration frames serve as a sample of the different possible track morphologies to test the efficiency and help the solidity of a machine learning model trained on simulations-generated images.

The corresponding papers will be linked below when published.