Fast Neutron Tomography of Fuel Cells Enabled By Seeded Tomography Reconstruction

Neutron imaging provides high sensitivity to liquid water while being able to penetrate materials typically used to construct fuel cells. Water distribution measurements in operating fuel cells are typically performed using neutron radiography as the neutron fluence rate of the beam requires long acquisition times for high resolution making tomography difficult. Radiography only provides a two-dimensional look at the water distribution with an operating fuel cell making tomography preferable to understand three-dimensional distribution of water. NIST developed a fuel cell hardware design to allow for simultaneous neutron and X-ray tomography and as first reported in 2017 [1], initial efforts for tomography required acquisition times of 18 hours while the cell continuously operated at steady state conditions. Progress was made to improve acquisition settings and reconstruction methods to reduce the time required. These improvements were presented in 2019 [2] and demonstrated that scan times could be reduced to 8 hours and provided ground work for reducing acquisition times down to 1 hour to 2 hours. Since 2019, further progress has been made in applying dose reduction iterative reconstruction algorithms to the fuel cell datasets to allow faster acquisition times. Faster acquisition scans were simulated by reducing the number of projection angles from the high-resolution datasets from the 2019 work. The high-resolution scan was acquired with 851 projections from 0° to 360° and the fast scan consisted of only 51 projections. Reducing the number of projections from 851 to 51 simulates a scan that requires only 29 minutes to acquire, which is close to the typical length of time for acquiring high resolution radiography images. Figure 1 shows the comparison of a single slice from the 3D volumes with (A) representing the 8 hour, 851 projection scan, (C) the iteratively reconstructed 51 projection scan, and (E) the 51 projection scan reconstructed with filtered back projection algorithm used for the 851 projection scan. The iterative reconstruction gives favorable results when compared to the 851 projection reconstruction and shows that flux limitations on speed can be overcome to achieve high-speed tomography at high resolution on a neutron beam. This talk will describe the methodology for acquiring neutron tomography scans of operating fuel cells in under 30 minutes and give insights into how this can be used towards solving water management issues. It will be p