Korecki and Materlik's method is novel because, unlike existing methods, it uses 'white X-rays' that cover a wide spectrum rather than X-rays of a single wavelength. It also images atomic structure in 'real space' - other techniques collect complex data about the phases of scattered waves, which must then be mathematically 'inverted' to obtain a real image.

X-rays are ideal probes of atomic structure because they have wavelengths comparable with the separation of atoms in crystals. But existing techniques based on X-rays are limited in resolution or application. Devices that focus X-rays through a lens suffer because the lens needs to be small - but this limits their resolution to about 10 nm, which is too coarse to resolve individual atoms. X-ray diffraction, on the other hand, are hindered by the so-called phase problem: the intensity of a diffraction pattern is independent of the exact position of an atom and this makes it impossible to convert the phase information into a real-space picture of the atomic structure.

The sample used by Korecki and Materlik was a silicon wafer 300µm thick. But the wafer also acts as a photodiode and uses the current produced to deduce the degree of X-ray interference at each point in the crystal. Combining this 'photocurrent data' with the direction of the incoming X-rays produces a two-dimensional image of the crystal structure. Korecki and Materlik slowly turned the crystal and collected eight such images, which they reconstructed to create the three-dimensional picture.

Korecki and Materlik also found that as the X-ray spectrum becomes broader - that is, as it includes a greater spread of wavelengths - the image becomes clearer. This is because the fainter fringes towards the edges of the diffraction pattern disappear as the range of wavelengths widens. Ultimately, only the central strong peak in the diffraction pattern - the 'zero-order' fringe - remains.

The new technique may allow physicists to establish exactly where dopant atoms are sited within crystals, which is not possible with existing methods. And because it makes direct imaging possible, single planes of the crystal can easily be scrutinized: existing devices require data from all angles of the crystal to form a real space image.