Exploring the duality of powder adhesion and underlying surface roughness in laser powder bed fusion processed Ti-6Al-4V

• Published in: Journal of manufacturing processes 2022-09, Vol.81, p.14-26
• Main Authors: Carter, Luke N., Villapún, Victor M., Grover, Liam, Cox, Sophie C.
• Format: Article
• Language: English
• Subjects: Image analysis; Laser powder bed fusion; Powder metallurgy; Surface roughness; Titanium alloys
• ISSN: 1526-6125; 2212-4616
• DOI: 10.1016/j.jmapro.2022.06.057

As-built surface roughness remains a significant barrier to widespread uptake of Laser Powder Bed Fusion (LPBF). To overcome this barrier, it is first necessary to understand the structure of this surface. This research explores the duality of the LPBF surface consisting of: 1) an apparent surface dominated by adhered powder and 2) the underlying profile defined by the fully consolidated material. An array of cuboidal specimens were produced by LPBF with varying contour scan parameters to study their influence on the vertical wall surface. For the first time, optical image analysis was used to quantify the extent and size distribution of surface adhered powder particles. SEM micrographs of the sectioned specimens were used in conjunction with a second novel image processing technique to study the underlying surface profile. Researchers utilised these methods to study the relationship between process parameters and both surface topographies. These results were compared against traditional line of sight roughness measurements. Surface finish (Sa), powder adhesion, and underlying roughness decreased with energy input to a point where the contour melt track was observed as unstable and incoherent (~0.1 J/mm). A 19.9 % reduction in Sa was demonstrated by the smoothest specimen (Sa, 10.9 μm) when compared to the control (Sa, 13.6 μm). Argon crossflow within the process chamber was shown to influence the underlying roughness, Ra, with upstream values 4.0 μm lower on average than downstream over most of the experimental range. Adhered particle size showed a finer distribution compared with the feedstock (D50 Surface = 20.8 μm; D50 Feedstock = 32.9 μm). Consequently, a further study comparing specimens built using coarse (sieved to >36 μm, D50 = 42.8 μm) versus fine (sieved to