XRM2024 - Mon02K - "Extended depth-of-field X-ray tomography at 4 nm resolution"

As ptychographic X-ray tomography 1 approaches single-digit or sub-1 nm 3D imaging resolution, it is the ideal candidate to bridge the resolution gap between transmission electron microscopy (TEM) and X-ray microscopy. This leap towards finer resolutions, however, introduces critical challenges: the depth-of-field dramatically narrows due to a square-law dependence 2, and the imaging process becomes increasingly susceptible to nanometer-sized instabilities. Therefore, imaging larger samples at high resolution requires new approaches in both image acquisition and reconstruction to expand the depth-of-field and correct for experimental errors. By addressing these challenges, X-ray nano-tomography can not only push the boundaries of resolution but also circumvent the intricate sample preparation required by TEM, while leveraging the high penetration depth of X-rays. This advancement would enable the detailed imaging of large samples necessary for connectome mapping of integrated circuits or brain samples, which is unattainable with electron microscopy, thus opening new frontiers in high-resolution X-ray tomography. To address these challenges and push the boundaries of high-resolution tomography, we introduce two novel methods aimed at mitigating X-ray beam instabilities and extending depth-of-field limitations beyond current capabilities. Our first approach, “burst ptychography”, involves capturing multiple low-exposure frames at each scan position to better resolve temporal dynamics of instabilities. This method simplifies the reconstruction process by reducing algorithmic complexity, compared to mixed-state ptychography 3 which relies on inferring dynamics from a single image. Secondly, we adopt a filtered back-propagation tomography algorithm from optical diffraction tomography 4 to image samples larger than the conventional depth-of-field. Our experimental results demonstrate this method's ability to extend the depth-of-field by at least tenfold without requiring multi-slice approaches 5, even for strongly scattering samples such as integrated circuits. We will showcase the performance of our proposed reconstruction algorithms through detailed tomographic reconstructions of integrated circuits, as well as other samples like a gyroid, down to 4 nm. Our presentation will discuss the novel algorithms and their potential to significantly advance the field of high-resolution imaging. References 1 Thibault, P. et al., (2008). Science, 321 (379–382).2 Tsai, E. H. R. et al., (2016). Optics Express, 24 (29089–29108).3 Thibault, P. et al., (2013). Nature, 494 (68–71).4 Kak, A. C., & Slaney, M. (2001). Principles of Computerized Tomographic Imaging. Society for Industrial and Applied Mathematics (203–273).5 Li. P et al., (2018). Scientific Reports, 8(1), 2049