Abstract
Electron tomography is an invaluable tool for characterizing the 3D structure of nanomaterials. Recently, there has been significant progress towards the real-time and fast approaches for determining 3D morphology of nanoparticles. These include quasi-3D imaging [1] morphology imaging with secondary electron beam induced current (SEEBIC)[2] and fast tomography.[3] However, these techniques on their own are insufficient for bulk analysis of material properties or characterization of samples with structures that alter rapidly such as beam sensitive MOFs and perovskites.
Herein we wish to investigate the use of these techniques in combination with computational techniques such as real-time neural radiance fields (NERFs), and machine learning[4] to perform real-time analysis of time sensitive nanomaterials. One such avenue would be to utilize fast tomography with NERFs to extract quickly extract a 3D scene which can be converted to a 3D volume for further analysis or to use machine learning with Fourier domain filters and multiprocessing to obtain real-time low resolution reconstructions for analysis while computing higher resolution reconstructions.
Researcher(s)
Research team(s)
Project type(s)