The core of GBI is the Talbot self-imaging effect: when spatially coherent X-rays pass through a grating (G1) with periodic line pattern, the line pattern will repeat itself at certain distances. By inserting a second grating (G2) that matches the self-image from G1, small differences introduced by a sample present in the beam can be detected. The second grating is shifted in several steps between where it is in phase and out of phase with the self-image from G1. This allows three properties of the object to be measured separately; X-ray attenuation, phase shift and dark field originating from ultra-small angle scattering. Separate attenuation and phase signals allow better discrimination of materials and the dark-field signal visualizes the presence of small structures below the imaging resolution.
By retrieving the changes in the self-image measured by stepping the G2 grating, the three image modalities can be measured simultaneously. GBI can often be implemented in a laboratory setup, but often requires a grating directly in front of the source, referred to as G0. The reason for this is that most sources don’t have the spatial coherence to enable image contrast in GBI, but each slit in the G0 grating works as a spatially coherent source but mutually incoherent to the other slits. Naturally, the G0 grating causes a loss in flux of more than 50%. A MetalJet X-ray source reaches the sufficient spatial coherence to enable GBI without the G0 grating, thus avoiding the loss of flux.
The most common way to acquire images in GBI is the phase stepping procedure that requires several images every projection. This has the downside that it associates GBI tomography to very long exposure times. In this context, it is of deepest concern to keep the flux high!