The in-line quality control and material recognition we develop spring from the same root: the physics of the interaction between X-rays and matter. Here we explain, with no shortcuts, how it works — and why doing it really well takes rare know-how.
X-rays are electromagnetic radiation, just like visible light or radio waves, but with much shorter wavelengths — typically between 0.01 and 10 nanometers — and therefore much higher energies, on the order of hundreds of electronvolts up to over 100 keV. It is precisely this high energy that gives X-rays their most useful industrial property: the ability to penetrate matter and to interact with it in ways that tell us what is inside an object and what material it is made of.
An X-ray photon carries an energy E linked to frequency by the relation E = h·ν (where h is Planck's constant): the shorter the wavelength, the more energetic — and penetrating — the photon. In practice, by tuning the X-ray energy you decide how deep to "look" and which materials to distinguish.
The fundamental distinction, also from a safety standpoint, is between ionizing e non-ionizing. Radiation is ionizing when its photons have enough energy to tear electrons from atoms, creating ions: this takes roughly more than ten electronvolts. X-rays and gamma rays are ionizing; so is part of the ultraviolet.
Instead, non-ionizing are visible light, infrared, microwaves and radio waves: they carry too little energy to ionize, and interact with matter mainly by exciting molecular vibrations and rotations (as is the case, for example, with microwaves). This difference has enormous practical consequences: working with X-rays requires shielding, dosimetry and radiation-protection procedures, because ionization can damage biological tissue — but it is also what makes it possible to "see" through metal and dense materials.
X-rays were discovered in 1895 by Wilhelm Conrad Röntgen, who observed a mysterious radiation able to pass through opaque materials and expose photographic plates; he called them "X" precisely because their nature was unknown. The famous radiograph of his wife's hand immediately made the diagnostic potential clear, and in 1901 Röntgen received the first Nobel Prize in Physics.
Today, X-rays for industrial use are produced mainly with X-ray tubes: electrons are emitted by a filament, accelerated by a high voltage (tens or hundreds of kV) and made to collide with a metal target (the anode, often tungsten). The impact produces X-rays through two mechanisms: braking radiation (Bremsstrahlung), a continuous spectrum produced by the deceleration of the electrons, and the characteristic lines, photons at very precise energies due to the inner electronic transitions of the target atom. For very high-brilliance applications, large storage rings (synchrotron sources) are used instead.
When an X-ray beam meets matter, it can behave in three main ways — and this is the key to understanding everything else.
Some of the photons pass through the object. The intensity that emerges follows the Beer-Lambert law, I = I₀·e^(−μx): the thicker and denser the material (and the higher its atomic number Z), the more it absorbs. This is the principle of radiography and tomography: the map of how much the beam is attenuated becomes an image of what's inside. Photoelectric absorption, dominant at lower energies, depends strongly on Z and on energy (it grows roughly as Z⁴ and falls as E³): this is why bone "shows up" more than soft tissue.
If an X-ray photon gives up energy by ejecting an inner electron of the atom, the atom "resettles": an electron from a higher level drops to fill the gap and emits a new X-ray photon of energy characteristic of the element. By measuring these lines you perform elemental analysis (X-ray fluorescence, XRF): you find out which elements the material is made of.
Some of the photons are deflected. In Rayleigh scattering (elastic) the photon changes direction but not energy; in Compton scattering (inelastic) it gives part of its energy to an electron and changes wavelength. Scattering organized by a crystal lattice gives rise to diffraction (XRD), which reveals the crystal structure of the material. Transmission, fluorescence and scattering happen together: the art lies in measuring the right signal and separating it from the others.
To turn X-rays into an image you need a detector. They fall into two broad families.
These are the classic flat panels: a scintillator converts X-rays into light, and an array of photodiodes accumulates the generated charge. They are robust and inexpensive, but they "sum" everything that arrives, mixing photons of different energies and accumulating noise: they cannot tell a weak photon from an energetic one.
I photon-counting detectors count individual X-ray photons and can measure their energy, applying thresholds. The advantage is enormous: almost zero noise, very high dynamic range and — above all — the ability to discriminate photons by energy, i.e. to do "spectral", color imaging. It is the leap that opens the door to material recognition.
Among photon-counting detectors, the Timepix (developed within the international Medipix collaboration, born around particle-physics research) represents the state of the art. They are hybrid-pixel detectors: a silicon sensor (or high-Z materials) coupled to a readout chip in which every single pixel is a small, independent measuring instrument.
The features that make them exceptional:
In short: a Timepix doesn't just take a grayscale "photograph", it collects where, when and with how much energy each photon arrives. It is a wealth of information that very few know how to fully exploit — and it is exactly what our architecture is built on.
A recurring question: if imaging and material recognition use the same detector, what's the difference? The difference is not in the hardware, but in the information extracted from the data.
Theimaging (radiography, tomography) builds a spatial map of attenuation: it serves to see the shape and internal defects — porosity, cracks, inclusions, weld defects. What matters most is spatial resolution and contrast.
The material recognition instead exploits the fact that attenuation depends on energy differently depending on the atomic number Z and the density of the material. By measuring how the signal changes at different energies ( dual-energy or spectral imaging), you obtain the effective Z of what you're observing: this is how you tell a plastic from an alloy, an organic material from an inorganic one, even if they have the same shape. Same spectral detector, same acquisition — but completely different processing: in imaging you look at the image, in recognition you analyze the spectrum pixel by pixel. This is where a detector like Timepix is at its best.
Putting this chain together — from source to detector to useful information — requires rare, deeply interdisciplinary expertise, of the kind formed in major research laboratories:
It is the combination of these skills — not a single piece of software or a single component — that makes the difference between "having a detector" and "knowing how to turn photons into industrial decisions".
We put this physics to work on your line: from the invisible defect to real-time material recognition.
Let's talk →