Radiation that does not play a part in conventional X-ray imaging has been exploited by physicists in the UK to provide comprehensive snapshots of an object's physical and chemical state. Potential applications of the new technique, known as "dark-field hyperspectral X-ray imaging", include identification of stress build-up inside engineered structures, security scanning of elicit materials, and analysis of medical biopsies.

Normal radiography of the kind used in hospitals relies on the phenomenon of absorption. A beam of X-rays is fired at an opaque object and the radiation that emerges on the far side is captured by a photographic film or digital detector, with the image mapping variations in electron density inside the object. However, the image cannot be used to identify the materials that make up the object in question.

Bright and dark fields

That limitation has been overcome in the latest work, which has been carried out by Robert Cernik of Manchester University and colleagues at Manchester and the Rutherford Appleton Laboratory in Oxfordshire. Instead of recording what is known as an X-ray beam's "bright field" – the radiation that passes through the sample – the new approach involves measuring a portion of the "dark field" – the radiation scattered or emitted by the object. "Usually great lengths are taken to remove the scattered radiation," says Cernik, "but in fact that radiation contains all sorts of extra information not available in conventional imaging."

The technique involves placing a sample in the path of a relatively wide polychromatic X-ray beam and then positioning a pinhole aperture a few degrees off the beam axis on the far side of the sample. A sensitive, multi-pixel detector then captures the radiation that emerges from the pinhole. Cernik explains that the set-up provides a new way of recording diffraction patterns from the sample. Conventional scattering experiments shine a monochromatic beam onto a crystal, which is rotated until the angle between the beam and crystal structure is such that the diffracted waves interfere constructively to produce a peak in output intensity. In the latest work, the sample and detector can remain fixed because each pixel is designed to record light intensity across a range of different wavelengths, producing what are known as "data cubes". With data from any one pixel revealing diffraction peaks at specific wavelengths, the combined output from all the pixels allows the various chemical elements and compounds that make up the sample, as well as their crystal structures, to be identified.

Imaging the sample – or identifying the positions of the various materials within it – therefore involves colour coding the output of each pixel according to which peaks it contains. This can be done not only for a flat, 2D sample, but also for real-life 3D objects. Rotating the sample by successive small amounts around a vertical axis and imaging it at each step generates a series of slices through the object that allows a "tomogram" to be built up in a similar way to the production of medical CT scans. The difference, as Cernik points out, is that in the case of hospital scans it is the detector and source that rotate, rather than the person being scanned. "Unlike humans," he quips, "our samples have no ethical rights."

Diamond devices

Cernik's group put its idea to the test using the Diamond synchrotron source – they placed a cylinder containing zinc oxide, aluminium and cerium oxide in the path of a square-shaped beam of white X-rays 8 mm wide for between two and five minutes at a time, and recorded the resulting images using a specially developed "High-Energy X-ray Imaging Technology" (HEXITEC) device.

Cernik told physcisworld.com that the technique could be used to monitor stresses inside objects, such as the complex welded components used in aircraft, with variations in crystal spacings across the object revealing any residual stresses. In addition, cracks in metals such as aluminium that are too small to image in any other way might be revealed by the corrosion that surrounds them – the diffraction pattern of aluminium oxide being different to that of aluminium.

Further, says Cernik, the ability to train a sensor to look for characteristic diffraction patterns could mean that the technique finds use in both security scanning – where the chemicals in question might be explosives or drugs – and in medicine. In the latter case, he explains, the technology might be best employed to reduce false-positive diagnoses; the non-identification of, for example, a certain kind of breast cancer within a biopsy helping to avoid unnecessary, unpleasant and costly treatment.

Quick scan

Cernik points out that 2D or 3D images showing the crystalline or chemical structure of complex opaque objects can already be carried out, thanks to the combination of X-ray diffraction or fluorescence together with tomography. But this approach is slow, he says, because it involves a narrow "pencil" beam of X-rays being scanned across or rotated around a sample to build up images bit by bit – as opposed to the direct imaging possible using his group's technique.

However, critics say that, while this latest work does have significant potential, the dark field generates very low intensities, making the experiment harder to carry out using a conventional X-ray tube. They argue that the team must develop a practical solution – such as using multiple pinholes – for the technique to be commercially useful.

In fact, Cernik has set up a company to commercialize his group's technology, and says that he and his colleagues are now working to reduce the cost and improve the spatial resolution of the semiconductor materials that make up the detector.

The research is published in Proceedings of the Royal Society A.