Search

This paper presents a deep learning algorithm for tomographic reconstruction (GANrec). The algorithm uses a generative adversarial network (GAN) to solve the inverse of the Radon transform directly. It works for independent sinograms without additional training steps. The GAN has been developed to fit the input sinogram with the model sinogram generated from the predicted reconstruction. Good quality reconstructions can be obtained during the minimization of the fitting errors. The reconstruction is a self-training procedure based on the physics model, instead of on training data. The algorithm showed significant improvements in the reconstruction accuracy, especially for missing-wedge tomography acquired at less than 180° rotational range. It was also validated by reconstructing a missing-wedge X-ray ptychographic tomography (PXCT) data set of a macroporous zeolite particle, for which only 51 projections over 70° could be collected. The GANrec recovered the 3D pore structure with reasonable quality for further analysis. This reconstruction concept can work universally for most of the ill-posed inverse problems if the forward model is well defined, such as phase retrieval of in-line phase-contrast imaging., (open access.)

Two in situ `nanoreactors' for high-resolution imaging of catalysts have been designed and applied at the hard X-ray nanoprobe endstation at beamline P06 of the PETRA III synchrotron radiation source. The reactors house samples supported on commercial MEMS chips, and were applied for complementary hard X-ray ptychography (23 nm spatial resolution) and transmission electron microscopy, with additional X-ray fluorescence measurements. The reactors allow pressures of 100 kPa and temperatures of up to 1573 K, offering a wide range of conditions relevant for catalysis. Ptychographic tomography was demonstrated at limited tilting angles of at least ±35° within the reactors and ±65° on the naked sample holders. Two case studies were selected to demonstrate the functionality of the reactors: (i) annealing of hierarchical nanoporous gold up to 923 K under inert He environment and (ii) acquisition of a ptychographic projection series at ±35° of a hierarchically structured macroporous zeolite sample under ambient conditions. The reactors are shown to be a flexible and modular platform for in situ studies in catalysis and materials science which may be adapted for a range of sample and experiment types, opening new characterization pathways in correlative multimodal in situ analysis of functional materials at work. The cells will presently be made available for all interested users of beamline P06 at PETRA III.