Development of an Additive Manufactured Artifact to Characterize Unfused Powder Using Computed Tomography

Abstract

Additive manufacturing (AM) is recognized as a core technology for producing high-value components. The production of complex and individually modified components, as well as prototypes, gives additive manufacturing a substantial advantage over conventional subtractive machining. For most industries, some of the current barriers to implementing AM include the lack of build repeatability and a deficit of quality assurance standards. The mechanical properties of the components depend critically on the density achieved. Therefore, defect/porosity analysis must be carried out to verify the components’ integrity and viability. In parts produced using AM, the detection of unfused powder using computed tomography is challenging because the detection relies on differences in density. This study presents an optimized methodology for differentiating between unfused powder and voids in additive manufactured components, using computed tomography. Detecting the unfused powder requires detecting the cavities between particles. Previous studies have found that the detection of unfused powder requires a voxel size that is as small as 4 μm3. For most applications, scanning using a small voxel size is not reasonable because of the part size, long scan time, and data analysis. In this study, different voxel sizes are used to compare the time required for scanning, and the data analysis showing the impact of voxel size on the detection of micro defects. The powder used was Ti6Al4V, which has a grain size of 45–100 μm, and is typically employed by Arcam electron beam melting (EBM) machines. The artifact consisted of a 6 mm round bar with designed internal features ranging from 50 μm to 1400 μm and containing a mixture of voids and unfused powder. The diameter and depth of the defects were characterized using a focus variation microscope, after which they were scanned using a Nikon XTH225 industrial CT to measure the artifacts and characterize the internal features for defects/pores. To reduce the number of the process variables, the measurement parameters, such as filament current, acceleration voltage, and X-ray filtering material and thickness were kept constant. The VGStudio MAX 3.0 (Volume Graphics, Germany) software package was used for data processing, surface determination, and defects/porosity analysis. The main focus of this study is to explore the optimal methods for enhancing the detection of pores/defects while minimizing the time taken for scanning, data analysis, and determining the effects of noise on the analysis.