We want to achieve unprecedented fast and accurate imaging of interfaces and faults in 3D semi-conductor materials, especially sub-22nm IC?s. To this end we will develop novel algorithms for globally-consistent element-specific tomographic reconstructions at near-atomic resolution based on multi-modal data from energy dispersive X-ray (EDX) and Scanning Transmission Electron Microscopy (STEM). Both modalities can be acquired simultaneously. We will exploit this multi-modal information in a completely new way, namely to steer all tomographic reconstructions in order to obtain one overall consistent 3D element-wise reconstruction. Improving each individual reconstruction in combination with image fusion of the tomograms from the electron and all X-ray channels, will deliver the ultimate tomogram by exploiting the strong aspects of every technique. We approach this new combined inverse problem by fusion of the information from the electron and all X-ray channels reconstructions. Use of priors in tomographic reconstruction reduces the number of tilts angles without loss of resolution. Dynamically adapting the resolution of projections and automatic assessment of the intermediate tomogram quality permits investigating dynamic acquisition schemes. This research will deliver near-atomic resolution of elemental and structural interfaces in tomograms with a reduced number of tilt angles and therefore enable fast, reliable fault detection for 3D IC?s.
