Objective Directed energy deposition (DED) is a popular additive manufacturing technology that uses a high-energy beam to melt metal powders and deposit them onto a substrate. It has the advantage of printing large-scale metal parts efficiently. Common high-energy sources for DED systems include laser, plasma arc, and electron beam. Laser DED (L-DED) is considered to print parts with better mechanical performance but low printing efficiency compared to plasma arc DED (PA-DED). Furthermore, there is a significant difference in the metallurgical mechanism between the two processes. In this study, we compared the geometry, microstructure, and mechanical properties of 316L stainless steel deposited by L-DED and PA-DED processes. The underlying mechanisms of the difference in geometry, microstructure, and mechanical properties of samples printed by the two processes were discussed.

Methods An in-house developed DED system that consists of a fiber laser with a maximum power of 6 kW, two robot arms, one L-DED module, and one PA-DED module was used. A 316L stainless steel powder with a particle size ranging from 60 to 125 μm was adopted as the feedstock for the two printing processes. Single tracks with a length of 80 mm were printed via the two processes, and their cross-sections were etched for geometry measurement. A quadratic regression orthogonal experiment was designed to investigate the effect of energy input, scanning velocity, and powder feeding velocity on the geometry of printed thin walls. The dimensions of the thin walls are 80 mm×3 mm×100 mm. The average layer width and height were measured from the middle location of the thin walls. The L-DED process parameters included a laser power of 2000 W, a scanning speed of 10 mm/s, and a powder feed rate of 24 g/min. The PA-DED process parameters included a current of 30 A, a scanning speed of 5 mm/s, and a powder feed rate of 12 g/min. Samples perpendicular to the build direction, parallel to the build direction, and inclined at 45° were machined for tensile testing. Microstructures of the printed thin walls were also observed from their cross-section locations.

Results and Discussions The average powder utilization rate of the two printing processes was calculated by measuring the weight difference before and after the deposition (Table 4). The average powder utilization rates of L-DED and PA-DED were 35.9% and 72.9%, respectively. The twice powder utilization rate of PA-DED compared with L-DED was attributed to the high-velocity of plasma arc that could accelerate powder particles deposited into melt pools during printing. The cross-sectional morphologies of the single tracks indicated that L-DED enabled a better metallurgical bonding between the melt pool and substrate than the PA-DED process. This was ascribed to the higher energy density of L-DED, increasing the penetration of the melt pool. The entire process was similar to deep penetration welding. In contrast, the small current used in PA-DED led to slight melting of the substrate surface, where the process was similar to conduction welding with a shallow melt pool. Therefore, a preheating process for the substrate or the utilization of a high current for the first printing layer should be conducted in the PA-DED process to enhance the bonding. The regression equations of layer width and height for PA-DED and L-DED were realized. The variance analysis of orthogonal experimental results (Tables 6 and 7) indicated that the process parameters of PA-DED and L-DED exhibited different influence orders on layer geometry. In the PA-DED process, the process parameters that influenced the layer width by the descending order were current, powder feed rate, and scanning speed, whereas the parameters that influenced the layer height by the descending order were powder feed rate, current, and scanning speed. Comparatively, in L-DED, the process parameters that influenced the layer width by the descending order were scanning speed, powder feed rate, and laser power, whereas the parameters that influenced the layer height by the descending order were powder feed rate, scanning speed, and laser power. The influence trend of process parameters on the geometry of the two processes was consistent when only a single factor was considered. As the energy input increased, the floor width increased and the floor height decreased; the layer width and height increased with an increase in the powder feeding quantity and decreased with an increase in the scanning speed. The microstructure morphologies (Fig. 5) of samples printed by the two processes were slightly different. PA-DED samples were dominated by directional growth long columnar grains with sizes up to millimeters, and many secondary dendrites could be obtained. However, L-DED samples showed shorter columnar grains and various growth directions in different regions. Particularly, the grains grew mainly perpendicular to the melt pool boundary. Tensile and microhardness testing results (Fig.6 and Fig.7) showed that samples fabricated by PA-DED achieved comparable mechanical properties to those printed by L-DED. The tensile strength of the samples was 593 and 570 MPa for L-DED and PA-DED, respectively.

Conclusions The powder utilization rate of PA-DED was significantly higher than that of L-DED. However, the dilution rate of the first layer of PA-DED was low on the substrate without preheating, leading to insufficient interfacial bonding strength. The prediction equations of the layer width and height of 316L thin-walled parts by L-DED were established. The effects of process parameters on the geometry of the two printing processes were compared. Current had a great influence on the layer width and height during PA-DED, whereas the influence of laser power on the layer width and height during L-DED could not be compared with the powder feed rate and scanning speed. The microstructure of PA-DED samples tended to grow directionally and their columnar grains were longer. L-DED samples obtained smaller columnar grains, which possessed various growth directions in different regions.