1. Introduction
Laser Powder Bed Fusion (L-PBF) is a widely adopted additive manufacturing process used to fabricate complex metallic parts. The process selectively melts a layer of metal powder by laser scanning across the powder bed. Subsequently, the powder bed lowers down one layer thickness, and a new layer of metal powder is spread on top of the melted layer. Again, the laser beam scans along the predetermined path, building the part layer by layer [
1]. The L-PBF process excels in producing complex geometries with single-step fabrication and reducing the weight of a component while meeting the mechanical requirements. These unique advantages have led to the process being utilized in various industries like biomedical, automotive, and tooling, but most notably the aerospace sector [
2].
Despite advances in the L-PBF process, it still faces challenges that hinder its growth in many industries. Due to lower dimensional accuracy and process repeatability compared to conventional methods, as-built parts require additional post-processing like polishing, coating, and chemical treatments. Moreover, mechanical properties, particularly fatigue strength, are constrained by factors such as porosity and surface roughness [
3].
One of the benefits of L-PBF comes from being able to additively build parts without the need for support structures. However, parts built with unsupported overhangs exhibit higher surface roughness, and the detrimental effect of surface roughness on fatigue performance is well examined. Sanaei and Fatemi have found that surface roughness significantly impacts fatigue performance, even in the presence of internal defects and varying microstructures [
4]. So, it is of high interest to explore ways to minimize surface roughness at low overhang angles. Also, surface roughness is an important consideration for L-PBF due to its effect on assemblability, which is defined as a component’s capability to assemble smoothly with a mating component and depends on the geometric tolerances of mating parts [
5]. In addition to considerations of friction, L-PBF-produced parts with high surface roughness require postprocessing steps, resulting in additional costs and time [
6]. Furthermore, some applications in the bio-medical sector require tailored surface roughness, so it is imperative that we establish the effect of process parameters on surface roughness [
7].
Among many ways to quantify and characterize surface roughness (S
a, S
sk, S
ku, S
t, S
q), the S
a value, which is defined as average height of a surface, is the most used as the other parameters have strong correlation to S
a value and overall provides a better estimate of surface roughness [
8]. To measure the roughness, contact stylus profilometry remains a staple for precise topography measurement at a microscale level. However, these contact-based methods are sometimes inadequate in measuring a surface area instead of along a line [
9]. Scanning Electron Microscopy (SEM) imaging allows for detailed visualization of surface features and provides valuable qualitative information about surface texture, morphology, and defects [
10,
11]. However, for precise surface generation with numerical modelling capabilities, X-ray Computed Tomography (XCT) and Scanning White Light Interferometry (SWLI) work well [
11,
12].
When examining surface roughness in parts built with unsupported overhangs, the surface facing downward (downskin) is found to have higher surface roughness than upward facing (upskin), vertical (sideskin), and horizontal surfaces [
13]. However, the causes and mechanisms of surface formation are different for upskin and downskin. Hence, the parameters affect both surfaces differently. Two primary contributors to upskin surface roughness are step edges and the agglomeration of partially sintered particles [
14,
15]. Strano et al. [
16] examined the relationship between overhang sloping angles and upskin surface roughness and found that for lower sloping angles, the staircase effect was the main cause of surface roughness, while at higher sloping angles, partially sintered particles attached to step edges was the dominant factor.
Downskin surface roughness is mostly due to the phenomena called dross formation which is caused by the difference in heat transfer rate between solidified metal and metal powder. As the laser irradiates a region with solid support, the heat dissipation is fast due to high thermal conductivity of solid metal. Conversely, when the laser hits a powder-supported region, the heat dissipation is slower and leads to larger melt pools. The enlarged melt pool then sinks into the powder bed due to gravity and capillary forces resulting in dross formation [
17,
18,
19]. The sagging melt pool produces a larger area for partially melted powder particle attachment, thus contributing to higher surface roughness.
A number of processes, design, and metallurgical factors affect downskin surface quality. Scan strategy (contouring, hatch spacing, skywriting) and process parameters (scan speed, laser power, hatch spacing, layer thickness), among others, impact downskin surface roughness. Tian et al. examined the influence of process parameters on surface roughness of Hastelloy-X parts made by L-PBF [
20]. They found that low laser power and high scan speed produce low downskin roughness. Fox et al. studied the downskin of stainless steel-GP1 parts at various overhang angles and observed overall surface roughness increase with decreasing inclination angle. Also, they found that partially melted particles dominate the downskin surface at a very low (0.06 J/mm) linear energy density, and at a higher (0.28 J/mm) linear energy density, along with partially melted powder, there are solidified melt tracks and deformed materials by the recoater on the surface [
21]. Snyder and Thole identified melt pool depth as the primary predictor of downskin surface roughness and found that contour parameters have significant impact for downskin surface [
22]. Ullah et al. identified sub-surface pores as another factor that affects the downskin surface quality of AM parts. The sub-surface porosity was found to increase with the inclination angle of the overhang surfaces. They also found that the contour scanning strategy can affect the downskin surface roughness as well as sub-surface porosity to a great extent [
23].
Although several studies have shown that contouring of the scan region is an effective way to reduce the surface roughness of L-PBF parts, a comprehensive analysis with various contouring methods and a wide range of contouring parameters still cannot be found. This study aims to study the effects of two different contouring strategies, namely pre- and post- contouring on the downskin surface roughness of sloped L-PBF parts. A wide range of contouring parameters including laser power and scan speed were taken into account, and the effects of these variables on downskin surface roughness were analyzed qualitatively and quantitatively from surface measurement results.
4. Discussion
The study reveals that both pre- and post-contouring strategies resulted in similar trends of surface roughness with the inclination angles and process variables. In pre-contoured samples, the lowest surface roughness of 14.08 µm was obtained for the 60°-inclined sample built with a low laser power (100 W) and high scan speed (2000 mm/s). The highest Sa of 42.84 µm was for the 30°-inclined sample with high LED (P = 195 W, V = 500 mm/s). The offset distance did not have a significant effect on the downskin surface roughness due to continuous remelting of the contouring region. Post-contouring using high LED for the inner contour and low LED for the outer contour, combined with a high scan speed, resulted in the low Sa of 18.88 µm for samples with a 60° inclination.
Irrespective of the strategy used, the downskin surface roughness is low at 30° inclination and it increases significantly for 45°- and 60°-inclined surfaces. The overhang distance between each layer is high for the 30°-inclined samples (51.9 µm) compared to the 45°- (30 µm) and 60° (17.3 µm)-inclined samples according to an equation proposed by Wang et al. (Equation (1)) [
28].
Here, S is the overhang length between two layers, H is the layer thickness, and θ is the inclination angle [
28]. The exposed melt pool region on the overhang causes dross formation and increases the surface area for partially melted powder particles. The surface skewness results showed that the peaks and valleys on downskin surfaces are not only process-induced but also due to random particle attachment (
Figure 5 and
Figure 11). In addition, the 30°-inclined surfaces showed a larger percentage area covered by particles (
Figure 8 and
Figure 12). The results align with Covarrubias et al.’s study on the effect of build angle on overhang surface quality. They also observed that the overhang surfaces built at smaller inclination angles have more visible partially melted particles, causing an increase in roughness [
13].
Likewise, the downskin surface roughness is high at pre-contouring with high laser power and low scan speed (high-LED). At high LED, the melt pool dimensions are higher than low-LED cases [
25]. This creates a larger melt pool area to be exposed to the powder bed underneath, aiding the attachment of partially melted particles to the surface. For post-contouring strategy, when both inner and outer contour scans had high laser power (P
i = 195 W, P
o = 150 W), the S
a value was higher too due to the large melt pool dimensions at high laser power. However, out of all the cases examined, the downskin surface roughness is lower when the outer contour laser power is low (100 W) and scan speed is high (2000 mm/s).
While comparing the surface roughness of samples built with similar contouring process parameters and same slope, the pre-contoured samples have lower roughness, even though the difference is merely ~5 µm. The selection of contouring strategies and processing conditions hence need to be tailored according to the desired end-use application. The effect of these strategies on other surfaces like vertical or upskin surfaces is another factor to be considered.
As an extension to this research, the upskin surfaces of fabricated samples were also analyzed [
29]. The effect of scan strategy and contouring parameters on the downskin surfaces of both pre- and post-contoured samples are quite distinct from their respective upskin surfaces.
Figure 15 compares the downskin and upskin surfaces of a post-contoured sample built with the medium-LED case (P
i = 150 W, P
o = 100 W, V = 500 mm/s) at low and high inclination angles. The upskin surface roughness is lowest for the 30°-inclined samples and it further increases with an increase in inclination angle. The upskin surfaces are mostly formed by the step edges, and the large distance between the step edges on 30°-inclined samples reduces their surface roughness [
25]. While using the higher scan speed and low laser power aids in reducing the downskin surface roughness, it negatively impacts the finish of the upskin region. A larger melt pool created with a low scan speed and high laser power helps the remelting on upskin surface and reduces the surface irregularities. Also, compared to the upskin regions where remelting happens on the previously solidified bulk regions, the melt pool size on the downskin surface will be larger for the same input energy density [
19]. The reason is the considerably lower heat conduction in the powder bed than the bulk regions because of the increased absorption of energy by powder particles during laser interaction [
26].
5. Conclusions
This experimental study evaluated the effects of contouring scan strategies and the process parameters on the downskin surfaces of sloped L-PBF parts. The influence of pre- and post-contouring strategies with a wide range of processing conditions like laser power, scan speed, offset distance, and inclination angle were analyzed. The WLI images and the average surface roughness, Sa, and surface skewness, Ssk, were used for qualitative and quantitative analysis. The major conclusions from the study are listed below.
Pre-contouring resulted in the lowest surface roughness (14.08 µm) for the 60°-inclined sample built with a low laser power (100 W) and high scan speed (2000 mm/s).
Post-contouring with a higher LED inner contour scan and low-LED outer contour scan was effective in reducing the roughness at higher scan speeds. The sample built with 195 W inner contour laser power, 100 W outer contour laser power, and 2000 mm/s scan speed has the lowest Sa value of 18.88 µm.
The pre-contouring at low LED resulted in overall smoother surface, but the difference in Sa values is very small (~5 µm) when samples built with similar conditions with both strategies are compared.
In general, a lower LED at the downskin surface–powder bed interface resulted in lower surface roughness, irrespective of the contour scanning strategy. The heat conduction through the powder bed is lower compared to the solidified bulk regions, and hence low heat input will help in smaller melt pool formation. This aids in reducing the powder particle attachment to the exposed area of melt pool.
The 30°-inclined samples have significantly higher downskin surface roughness compared to the 45°- and 60°-inclined samples. The overhang length is higher at smaller inclination angles, which results in more powder particle attachment to the downskin region.
The conditions favorable for the lower downskin surface roughness have a negative impact on the corresponding upskin surfaces. A high LED helps in remelting the upskin surface which results in better surface finish. Also, the upskin surface roughness reduces with an increase in distance between step edges, and hence 30°-inclined samples have smoother surfaces.
The contouring of the scan domain is an effective method to reduce the downskin surface roughness of L-PBF parts. The selection of contouring scan strategy and processing conditions depends on the end-use application. However, this study provides a comprehensive approach to evaluating the process variables to tailor the downskin surface roughness of L-PBF parts. The careful consideration of contour scanning strategy and processing conditions can help in attaining a desired surface quality of the parts by eliminating the post-processing stages.