Abstract
Reactor-based neutron radiography is a non-destructive, non-invasive characterization technique that has been extensively used for engineering materials [1] such as inspection of components [2, 3], evaluation of porosity, and in-operando observations of engineering parts [4]. Neutron radiography has flourished at reactor facilities for more than four decades and is relatively new to accelerator-based neutron sources [5]. Recent advances in neutron source and detector technologies, such as the Spallation Neutron Source (SNS) at the Oak Ridge National Laboratory (ORNL) in Oak Ridge, TN, and the microchannel plate (MCP) detector [6-8], respectively, enable new contrast mechanisms using the neutron scattering Bragg features for crystalline information such as average lattice strain, [9] crystalline plane orientation, and identification of phases in a neutron radiograph [10]. Additive manufacturing (AM) processes or 3D printing have recently become very popular and have a significant potential to revolutionize the manufacturing of materials by enabling new designs with complex geometries [11] that are not feasible using conventional manufacturing processes. However, the technique lacks standards for process optimization and control compared to conventional processes. Residual stresses are a common occurrence in materials that are machined, rolled, heat treated, welded, etc., and have a significant impact on a component’s mechanical behavior and durability. They may also arise during the 3D printing process, and defects such as internal cracks can propagate over time as the component relaxes after being removed from its build plate (the base plate utilized to print materials on). Moreover, since access to the AM material is possible only after the component has been fully manufactured, it is difficult to characterize the material for defects a priori to minimize expensive re-runs. Currently, validation of the AM process and materials is mainly through expensive trial-and-error experiments at the component level, whereas in conventional processes the level of confidence in predictive computational modeling is high enough to allow process and materials optimization through computational approaches. Thus, there is a clear need for non-destructive characterization techniques and for the establishment of processing- microstructure databases that can be used for developing and validating predictive modeling tools for AM.