Effects of Build Angle on Additively Manufactured Aluminum Alloy Surface Roughness and Wettability

The drive to create more powerful, yet increasingly small energy sources has led to excessive heat fluxes in electronics packaging. High heat transfer within a concentrated area can result in electrical components overheating and/or warping. One solution to this problem is the use of a multiphase heat spreader (i.e., thermal ground plane) such as a vapor chamber (VC) that effectively spreads heat within a hermetically sealed hollow chamber (or cavity) using a working fluid and wicking structures [1,2]. Vapor chambers utilize the fluid phase change to create temperature and pressure differentials, thereby promoting the liquid and vapor slug movement throughout the cavity [3]. The design of a VC is somewhat limited in its geometrical complexity due to limitations in conventional manufacturing (e.g., end-milling, casting) and the need to integrate vacuum-grade joints to ensure hermetic sealing. In addition, wicking structures, which may exist as microchannels, sintered particles, or meshes, require cumbersome integration methods. With the use of layer-by-layer fabrication offered through additive manufacturing (AM) methods, such as laser powder bed fusion (LPBF), VCs with more complex geometric features are now possible [4,5].

Laser powder bed fusion is an AM process commonly used for “printing” metals. It uses a concentrated laser for selectively melting and fusing regions of a pre-deposited powder bed [6]. A thin layer of powder (e.g., 30 µm) is evenly spread across a build plate; subsequently, the laser melts select regions corresponding to a solid model path, and once the entire cross-sectional area of the part has been fused, another layer of powder is evenly applied on top of the previous layer. This process is repeated until the part has been completed [7].

There are several design parameters associated with LPBF. The orientation and the location of the part during printing will influence the quality (e.g., roughness, porosity, geometry, mechanical properties) of the final part [8–10]. One such study observed that the printing orientation of LPBF AlSi10Mg deviated the R_a value of surfaces by as much as 5 µm, thus in some cases, creating stress concentrations leading to an over 50% reduction in fatigue strength [11]. Furthermore, unmelted powder within any channels and/or cavities within the part must be removed after LPBF.

The benefits of using AM to produce complex VC geometries include being able to design and manufacture optimized and/or conformal designs with as-printed wicking structures more easily and cost effectively. In the study by Liu et al. [12] on the performance of VCs designed to mimic plant leaves, the process used to create the complex geometries ranged from chemical etching to graphite molding. The channel and wick designs employed were found to significantly improve heat transfer efficiency; however, the time and cost to produce such designs utilizing traditional manufacturing make scaled production of such VCs challenging. An experimental investigation was performed on a loop heat pipe with an additive-manufactured stainless steel primary wick cooling an 80 w streetlamp. The heat transfer rate of the AM wick was 10% higher than a conventional wick loop heat pipe [13]. A porous regularly patterned stainless steel wick structure created using selective laser melting technology was studied for its effective thermal conductivity, contact angle with multiple fluids (i.e., water, hexane, and ethylene glycol), and capillary performance. The wick showed improved capillary performance compared to sintered, mesh, or composite wicks; the large permeability was noted as the reason for the improved performance [14,15]. A study conducted on the design of cooling layers for manufactured tooling determined that producing complex lattice structures via AM reduced cooling times within the cooling layers by 26% over a comparable design made utilizing traditional manufacturing methods [16]. Another study conducted on the feasibility of producing an oscillating heat pipe with Ti-6Al-4V found a 500% increase of thermal conductivity over a solid titanium alloy [17].

Within a heat pipe, it is important to understand the surface roughness and the wetting phenomena because of their direct effects on heat transfer. For example, the effects of surface roughness during natural convection play a major role in heat transfer. Du and Tong demonstrated that modifying the surface from a “smooth” to a “rough” surface with 9 mm pyramids improved the circulation within a fluid chamber such that it increased thermal transport by 76%. In addition, the change in surface roughness affected the number of “thermal plumes,” which were found to enhanced heat transport [18].

The surface roughness also impacts the contact angle. The contact angle of a droplet affects the surface area of the liquid on a solid surface, which in turn affects the rate of heat transferred to the liquid. The rate of evaporation, and therefore heat transfer, has been found to change depending on how a liquid droplet interacts with a surface. Contact angles directly affect the thin film thickness of the liquid which is where the majority of evaporation heat transfer occurs [19]. For water, contact angles that are less than 90 deg (i.e., hydrophilic) generally have linear evaporation rates, whereas contact angles that are greater than 90 deg (i.e., hydrophobic) typically have nonlinear evaporation rates [20]. The contact angle of water on AM AlSi10Mg is considered to be hydrophobic [21]. Contact angles in the context of VCs are mostly affected by heat transfer rate, solid heat transfer properties, liquid/gas heat transfer and fluid dynamics properties, surface roughness, internal surface geometries, and surface treatment. A droplet sitting on a surface will be in one of the two states—Cassie or Wenzel. In the Cassie state, the droplet sits on top of the microscale ridges of the surface typically on pockets of air; in the Wenzel state, the droplet sits inside the microscale ridges of the surface and is pinned to the surface. Typically, when a surface is hydrophobic, increasing surface roughness will make the droplet more hydrophobic, and when a surface is hydrophilic, increasing surface roughness will make the droplet more hydrophilic. Both hydrophobicity and hydrophilicity have their advantages for heat transfer. In condensation, a hydrophobic surface sheds droplets faster to keep the liquid film resistance lower, thereby increasing heat transfer [22,23]. In boiling, a hydrophilic surface is advantageous to maintain a layer of liquid water on the heated surface, thereby minimizing the effect of dryout [24].

In general, the LPBF process can result in highly variable surface roughness [25] and porosity due to the nature of the melting and solidification processes [26]. Processes can be performed to change surface conditions with coatings [27], remelting [28,29], and adding microstructures [30,31], which can affect fluid contact angles. Parts made via LPBF are prone to having a surface roughness that varies along their perimeter—based on how that perimeter is oriented with respect to the build direction [32]. Surface portions that are facing upward are “upskins,” while surface portions facing downward, toward the substrate, are “downskins” [25]. Upskin surfaces are typically smoother due to there being more solidified material underneath them and gravity pushing the melt pool to spread along the surface more, while downskin surfaces are typically rougher as they interact more with the powder bed where capillary forces pull on the melt pool during processing.

Several studies have focused on understanding the process–structure–property relationships inherent to the LPBF of AlSi10Mg; in particular, how process factors such as laser power, laser beam diameter, layer thickness, and build orientation affect AM surface quality [33–37]. Although one can optimize the specific energy of the laser during LPBF to reduce surface roughness [33,34], one can also control surface roughness via part orientation. Even if one does not aim to control final part roughness, it is beneficial to understand how different oriented surfaces vary in terms of surface roughness, especially for those intending to use LPBF parts for thermal/fluid applications. By using a laser scanning microscope, Leis et al. [33] measured an effective surface roughness of 22.3 µm for the surface of each AlSi10Mg powder layer in their LPBF system. It is mentioned that the rough powder layer provides for variation in local absorptivity and thus uneven laser-to-part energy transmission. Li et al. [34] found that the surface roughness along the top face of their LPBF AlSi10Mg specimens was far less that that measured along the sides. However, Yu et al. [29] provide results demonstrating that the sides of their AlSi10Mg LPBF cubes were smoother than their top surface. This may be attributed to Yu et al. preheating the substrate to 200 °C on their SLM Solution system; yet, this also demonstrates how the variability in LPBF system can lead to unique surface roughness observations. Dong et al [35]. demonstrated that the net shape is closer for AlSo10Mg LPBF specimens (flat tensile bars) built at an angle relative to vertical specimens. Hofele et al. [36] investigated the effect of build angle on surface roughness of LPBF AlSi10Mg specimens and found that surface roughness generally decreases when going from a 0 deg (horizontal orientation) to a 60 deg build orientation. For the 45 deg-oriented specimens, the downskin surface roughness was found to be nearly double that of the upskin. Arithmetic roughness (per ISO 4287) ranged between 8 and 19 µm. A maximum height variation of 170 µm was observed for the 0 deg sample due to the staircase effect. Gouveia et al. [37] printed various test artifacts from AlSi10Mg powder via LPBF. They found that the typical average roughness was around 8–12 µm and that regions with no support structure could possess up to 20% higher surface roughness. They also found higher variability in surface roughness measurements for top surfaces due to intermittent defects relative to side surfaces. The roughness of the top surface was found to be slightly higher than that of the sides, with this difference becoming essentially zero when using the manufacturer-provided, default process parameters. This study aims to determine the effects of printing orientation surface roughness and fluid contact angles for as-printed LPBF AlSi10Mg surfaces. The relationship between the LPBF AlSi10Mg surface and surface wetting has not received significant attention in the literature. This study observed uncoated, untreated surfaces, which would naturally be present inside of a heat pipe with no surface postprocessing, since in most heat pipe applications, the inclusion of surface coatings and treatments would involve expensive postprocessing that is time consuming and results of varying benefits. The study of different surface coatings and treatments in AM AlSi10Mg heat transfer applications may be beneficial for future studies. The information in this study will aid in the design considerations of printing orientation on multiphase thermal heat spreaders. This study also seeks to provide insights on surface roughness and fluid wetting behavior that will be present in such multiphase thermal heat spreaders.

Specimens were fabricated via LPBF using spherical AlSi10Mg (cast aluminum alloy) powder (i.e., 10–45 µm normally-distributed diameters) as the raw material. This specific alloy was chosen due to its additive manufacturability, low cost, and relatively high thermal conductivity [38]. Although copper is the typical material of choice for thermal management media, it is a considerably more expensive material to manufacture via LPBF. It is also a much denser material making it less viable in applications requiring low-weight thermal solutions. Currently, the processing of copper alloys via LPBF is still in development and not widespread. Hence, the use of LPBF of aluminum thermal management devices is of interest and the focus of this current study. The LPBF of all specimens were fabricated using LPBF services provided by Protolabs. Each layer was set to a thickness of 30 µm, providing for high resolution with tolerances of ±0.1 mm. The specimens were manufactured with a GE Concept Laser M2 using a 370 W Nd:YaG fiber laser with a spot size of 200 µm and a scan speed of 1800 mm/s. The material composition provided was as follows: Al (87.1–89.35 weight %), Si (9–11%), Fe (0.55%), Cu (0.05%), Mn (0.45%), Mg (0.2–0.45), Ni (0.05), Zn (0.1), Pb (0.05%), Sn (0.5%), and Ti (0.15%).

Seven specimens with dimensions shown in Fig. 1 were printed, each with its own build orientation/angle, ϕ (i.e.,: 0 deg, 10 deg, 20 deg, 30 deg, 40 deg, 50 deg, and 60 deg) normal to the build plate surface. All specimens had the same design consisting of 5 mm tall pillars with 5 mm diameters for the potential use inside a VC. Such pillars provide chamber support and another means for mass and heat transfer via wicking [39]. Five pillar geometries with chamfer angles of 30 deg, 25 deg, 20 deg, 15 deg, and 10 deg were created to determine if there were significant effects on surface roughness with a change in the printing angle. The specimens were printed with the flat surface opposite to the pillars as the base, meaning the pillared surface was the “upskin” of the part during LPBF. Apart from the pillar geometries, there was space on the upskin of the specimen, which was intentionally left flat to measure the surface roughness and contact angles, which may represent VC surfaces within the cavity. As shown in Fig. 1, when changing the build angle, the orientation angle of the pillars changed. This effect on the upskin and downskin (identified in Fig. 1) surface roughness was investigated through microscopy of the pillar profiles.

Before any measurements, the specimens were rinsed with isopropyl alcohol, acetone, and water and then allowed to dry. This cleaning process represents one used to clean and remove excess, entrapped metal powder from within a cavity of a VC manufactured via LPBF. Specimen surfaces were not coated to better represent the surfaces of as-printed internal geometries. The typical procedure for the removal of excess metal powder within an internal geometry is to implement sealable ports into the side of the specimen, which enables it to be rinsed [17].

To characterize the relationship of surface roughness and contact angles with build orientation of the AlSi10Mg specimens, two experiments were conducted. The first experiment aimed to determine the solid–liquid, initial contact angle, θ, with a 0.2 µL (± 0.08 µL) deionized water droplet on the flat portion of the upside of each specimen. The contact angle of the water droplet was measured using a digital goniometer (Allied Prosilica GC model # GC750 using the software FTA 32), the contact angles were measured using photon fastcam viewer 4, and the results were exported to a microsoft excel^® spreadsheet for the analysis (Fig. 2). The droplet was placed on the specimen surface a Fisher Brand Elite Adjustable-Volume pipette oriented as vertical as the goniometer allowed; the surface temperature was monitored using a type-T thermocouple (±1 °C resolution) resting on the same sample surface as the droplet. Contact angle measurements were carried out at room temperature, nominally 22 ± 1 °C.

Once the 0.2 µL water droplet had ceased moving from being placed on the specimen’s surface, achieving a steady-state morphology, a picture (640 × 480 96 dpi) was taken, which enabled the solid–liquid contact angle to be calculated by the digital goniometer software. The droplets remained in a steady-state morphology for several minutes on the surface of the specimen at room temperature before completely evaporating. The contact angles of three water droplets placed in different locations on the flat surface of each specimen were measured by the digital goniometer, for a total of 42 contact angle measurements. The contact angles of both the left and right sides of the two-dimensional water droplet image were obtained, and an average contact angle calculated from both the left and right contact angles was obtained from each droplet. The average contact angle of all three water droplets on each specimen was also calculated. An example image of a droplet with contact angle measurements obtained via the goniometer is shown in Fig. 3. Note that the small weight of the droplets prevented any gravity influence on droplet morphology as shown by its Bond number, Bo, being much less than unity (discussed later).

The second set of experiments focused on determining the impacts of build orientation and pillar geometry on surface roughness. Several images of each specimen were taken utilizing an optical microscope (Amscope ME520 T) magnified 5×. Side and top view images were taken of each flat surface, while only side-view images (at two different magnifications) were taken for each pillar. The sides of each pillar were observed with the optical microscope to determine upskin/downskin variation and whether a significant difference in surface roughness was present between chamfer angles on the same specimens. The pillar roughness profiles are important since, once incorporated into the internal geometry of a multiphase heat spreader, they act as wicking structures. Since a majority of evaporation takes place in the bridging area near the solid–liquid interface, the surface of these structures implies the wetting behavior of liquid droplets; and therefore, heat transfer. The flat surfaces were important to observe to determine whether a trend between printing orientation and surface roughness, and consequently, liquid contact angle, was present. Figure 4 shows the pillar geometry and location of the pictures taken with an optical microscope for the specimens throughout this study.

Once images had been taken using the optical microscope, the specimens were cut into six pieces so that they could be observed under a scanning electron microscope (SEM). SEM microscopy was performed using a Low Vacuum SEM (Model: Carl Zeiss EVO MA 10) operating at 5 kV. Due to the specifications of the microscope, the specimens could not fit into the observation platform without first being separated. Each of the pillars, as well as the flat portion of the specimens, was separated from each other and rinsed to remove contaminants.

Specimens were then measured for surface roughness at five locations via an EDM MVi5 profilometer with FANUC 5-Axis software. A roughness measurement was taken along the flat surface of each part in the same location that contact angles were previously recorded. Two different roughness measurements were taken on both the left (downskin) and right (upskin) sides of each 10 deg chamfer pillar. Theoretically, on the 0 deg-built specimen, the downskin and upskin surface roughness values would be equivalent because their build orientation would be the same. One measurement was taken in a location with a large partially melted metal bead to obtain a value of high surface roughness, and one measurement was taken in close proximity to this location such that there were relatively few partially melted metal beads. This was done on both the upskin and downskin sides of the pillars to obtain high and low surface roughness values to provide a more accurate surface characterization.

A total of 42 initial contact angle measurements were recorded for the flat surface of each specimen, and the results are presented in Table 1. To determine whether the values obtained in this study were statistically relevant, the parameter p-value was calculated using a single factor analysis of variance. The p-value was determined to be 0.001, which is less than the value 0.05 required to verify that the contact angle values obtained in this study were statistically relevant. AlSi10Mg alloy is typically hydrophobic [21]. However, data presented in Table 1 and Fig. 5 indicate that the build angle, ϕ, influences wettability and subsequent contact angle, θ. Variation was observed between the left-hand-side and right-hand-side contact angles (i.e., anisotropic wetting) for some droplets, indicating some nonuniformities in surface roughness and composition due to the LPBF manufacturing process. For build angles of 0 deg and 60 deg, contact angles indicate that the surface is primarily acting hydrophilicly although the variance is larger for the 0 deg build angle. At build angles of 10 deg and 20 deg, some droplets were hydrophilic on one side and hydrophobic on the other side; it is possible that partially melted metal beads, as discussed in the following section, may contribute to this nonuniformity. For 30 deg, 40 deg, and 50 deg build angles, droplets consistently exhibited hydrophobic behavior and partially melted beads were smaller in diameter. It was observed that there was a higher amount of partially melted metal beads on the surface of the ϕ = 0 deg specimen; upon contact with partially melted metal beads left over from the LPBF process, the water droplet would “attach” itself via surface tension, which greatly affected the contact angles [4,28,40]. As the printing angle increased, the surface became more water repellant until the experiments reached the fifth specimen (ϕ = 40 deg). It was observed that anisotropic wetting (i.e., left-hand-side and right-hand-side variation) generally decreased as the build angle increased. These trends indicate that the surface roughness decreases as the build angle increases, which allows for more droplet stability.

Per Young’s equation, it is clear that roughness directly impacts contact angle. For a unity roughness, i.e., γ = 1, the contact angle is equivalent to that experienced on a smooth surface. For hydrophilic surfaces, roughness often decreases contact angle, whereas for hydrophobic surfaces, roughness can increase the contact angle [41].

Via optical microscopy, it was observed that the sides of the pillars along all seven specimens contained a considerable amount of partially melted beads. Per Ref. [26], the ratio between laser power and laser beam diameter, P_L/d_b, used in the LPBF of AlSi10Mg parts was optimum in the range of 3000 W/mm < P_L/d_b < 3500 W/mm. Values lower than 2000 W/mm generally led to the creation of partially melted metal beads and uneven roughness values [26]. For this study, P_L/d_b = 1850 W/mm, and this explains the excessive formation of partially melted metal beads on the specimens in this experiment. However, in the context of VCs, a large surface area and a rough texture may in fact benefit heat transfer and bubble formation during the fluid phase change [42]. Images of the partially melted metal beads on the pillars are shown in Figs. 6 and 7.

In Fig. 6, the shared downskin/upskin sides of pillar 1 are visible. These are the sides that are visualized when looking directly at an upskin or downskin. It is interesting to note that the shared downskin/upskin sides are smoothest for the 40 deg and 60 deg build orientations, while roughest for the 30 deg and 50 deg build orientations. For Fig. 7, note that Pillar 1 has a chamfer angle of 30 deg and pillar 5 has a chamfer angle of 10 deg, with each pillar chamfer angle in between varying by 5 deg (see Fig. 1(a)). From Fig. 7, it may be seen that partially sintered beads possessed widths that varied from ~20 µm to as high as 174 µm. Agglomerations of powder with satellites are evidenced in Figs. 7(a), 7(d), and 7(e), indicating that the chamfer angle had little-to-no influence on agglomeration. There is no clear trend between pillar chamfer angle and surface roughness. However, it does appear that the steepest chamfer angle of 10 deg provided for the smoothest surface with the exemption of the lone agglomeration along its right side. All other chamfer angles provided for pillars with profound agglomeration—more pronounced and sporadic than that observed for pillar 5 (10 deg chamfer angle).

The flat upskin surface (nonpillared) of each part was observed from two different angles, from the side and from the top (Figs. 8 and 9, respectively). A trend in partially melted bead size was observed; as the printing angle increased, the bead size generally decreased. Partially melted beads were most prominent for low build angles and diameters decreased with the increasing build angle. This, in the context of how the specimens were printed, makes sense as the effect of gravity would pull these beads to the side, rather than leave them standing upright [43]. For the horizontally printed specimen, the flat upskin was found to have one of the largest attached particles in this study, at 234 µm. The surfaces without partially sintered particles were found to have numerous “mounds,” indicating some type of balling effect taking place during LPBF. Figure 9, which is a top view of the flat portion of the specimens, continues to show how gravity perpendicularity affects the LPBF melt pool and subsequent surface features. Surface roughness variability, as well as a few partially melted metal beads, which have formed because of the AM process, can be seen in these figures. The flat surface of specimen 5 (ϕ = 40) was the smoothest. It had the smallest difference in height and had relatively fewer partially melted metal beads. This corresponds directly with the contact angle observed earlier and explains the trend in increasing water repellency as the printing angle approached 40 deg. Parallel track lines are observable in Fig. 9; however, they become less apparent as the build angle increases, especially for those greater than 40 deg. Any light contrast present in Figs. 8 and 9 is the result of the lighting required to obtain high resolution images of the specimens.

As shown in Fig. 10, profilometer measurements were performed to obtain five average roughness values, R_a, along various locations of the specimens. One measurement was obtained from the flat upskin portion of the part in the same location that the contact angle measurements were previously obtained (Fig. 11) and four measurements were obtained along the sides of the pillars (Fig. 12). The pillar roughness measurements were taken on both the left and right sides of the pillar corresponding to their upskins and downskins during LPBF, respectively. Note the evidence of localized burning along the pillar shown in Fig. 10 occurring along a shared upskin/downskin side.

As shown in Fig. 10, the upskin profilometer measurements were averaged over a path of 8.25 mm in length, while the pillar profilometer measurements were averaged over a path of 4.15 mm in length along their height. From Fig. 11, the upskin (left side) surface of each pillar was found to possess average roughness values between 8.04 and 12.4 µm, whereas the downskin (right side) surface had average roughness values between 9.25 and 17.6 µm. The upskin and downskin surface roughness of the investigated pillars clearly depend on the build angle. In all cases, and as expected, the upskin surface of the pillar was smoother than its downskin surface. This is attributed to the downskin being surrounded by unmelted powder, which acts as a thermal insulator and wicking medium, causing the melt pool to run hotter and be “pulled” upon, respectively. The upskin has the advantage of having solid metal beneath it to help conduct heat away. In addition, the gravity force depresses the melt pool more due to the underlying solid preventing any permeation, generally resulting in smoother profiles. Significantly more variance in roughness was found for the pillar downskin relative to its upskin. A maximum roughness near 18 µm was measured for the pillar downskin on the 50 deg-built specimen. The 0 deg, 10 deg, and 60 deg-built specimens possessed pillars with downskin roughness values between 12 and 14 µm. The smoothest pillar downskins were found for the 20 deg, 30 deg, and 40 deg-built specimens, ranging between 9 and 12 µm. The upskin surface roughness values of the pillar generally increased with specimen build orientation. The upskin surface was smoothest for the 0 deg and 10 deg build orientations with values near 7–8 µm. The results demonstrate an inflection point in upskin and downskin surface roughness values—occurring around a specimen build orientation of 50 deg. The upskin surface roughness of the specimen flat region was highest for the 0 deg build orientation (~16.5 µm) and generally became smoother as the build orientation increased, being approximately 5 µm at the 60 deg build orientation. The trend is broken for the 50 deg build orientation, in which the upskin surface roughness of the specimen’s flat region is higher than that for the 40 deg build orientation. The upskin surface roughness barely changed for build orientations between 20 deg and 40 deg.

The roughness values for notable partially attached beads (defects) on the upskin (left) and downskin (right) portions of the investigated pillar are shown in Fig. 12. It may be seen that there is strong variance in defect roughness, ranging between 11 and 38 µm for the pillar upskin and 12–27 µm for the pillar downskin. The defect roughness appears to have no relationship with build orientation due to the random powder size distribution used during LPBF. The variance in defect roughness is less for the pillar downskin, and this may be attributed to powder bed capillary effects.

The nondimensional Bond number, Bo, represents the importance of gravitational forces to surface tension forces:

A nominal diameter of 1.38 mm was obtained for a contact angle of 40 deg and a diameter of 0.75 mm for a contact angle of 110 deg; compared to the roughness, maximum peaks of 204 µm were observed on the surface with a build angle of 50 deg. Roughness impacts wettability. According to the classic theory by Wenzel, for wetting surface (i.e., θ < 90 deg), increased roughness increases hydrophilicity, while for nonwetting θ > 90 deg, increased roughness increases hydrophobicity [48]. In this study, the contact angle of 90 deg distinguishes between the hydrophobic and hydrophilic regimes, and the AlSi10Mg surface is hydrophobic by nature [21], so according to the classic Wenzel theory, hydrophobicity should be observed. However, Wenzel’s theory does not adequately predict behavior for anisotropic rough surfaces. Plawsky et al. [49] note that a sessile liquid droplet placed on a rough surface may not be in either a Wenzel state (i.e., liquid fully penetrates roughness) or Cassie-Baxter state (i.e., air pockets in the roughness support the droplet). Droplet behavior depends on the interaction between the liquid and rough surfaces. Liquid may partially imbibe in the rough surface, and droplets on the size of the roughness may have varying contact angles around the droplet’s perimeter. Huang and Gates [50] observed 5 µL water droplets on rough surfaces (e.g., hydrophobic sand, glass beads, and a combination of particles). Contact angles varied based on the perimeter in contact with the rough surface.

The LPBF process results in heterogeneous surfaces, as demonstrated by the partially melted beads shown in Fig. 8, microscopy images presented in Figs. 7, 9, and 10, and the surface characterizations shown in Figs. 14 and 15. Wetting behavior depends on the surface energy of the material, nanoscale and microscale structures, and periodicity of the structures [51–56]. Natural superhydrophobic surfaces such as butterfly wings rely on asymmetric microstructures or nanostructures, thereby resulting in anisotropic wetting [51]. Engineered surfaces can also promote anisotropic wetting. Wang et al. [52] investigated wetting on hydrophilic and hydrophobic stripe patterns, observed nonhemispherical equilibrium droplet shapes, and noted “the anisotropic wetting makes Cassie−Baxter theory no longer applicable globally.” Similarly, Vrancken et al [53]. investigated wetting on polygonal posts, which resulted in polygonal shaped droplets. They expressed the importance of the lattice arrangement and shapes of the posts in dictating anisotropic wetting.

The results shown in Figs. 14 and 15 indicate that the lack of periodic structures (e.g., partially melted beads observed in Fig. 8) atop the LPBF-manufactured surfaces impacts wetting and can lead to anisotropic wetting (i.e., hydrophilic and hydrophobic behavior in the same droplet). The build angle of 0 deg has one of the highest values of Sa and Sdr and also corresponds to the differing contact angle behaviors, with most contact angles hydrophilic, yet one droplet experiences anisotropic wetting, in which the left side of the droplet is hydrophilic (i.e., θ = 81.7 deg) and hydrophobic on the right side (i.e., θ = 93.6 deg). Arithmetic mean heights and developed interfacial area ratios are the lowest for build angles of 30 deg and 40 deg, which maintain their hydrophobicity for all tested droplets, indicating that the more consistent surface roughness may be responsible for the hydrophobic contact angles. Belaud et al. [54] noted that although r, the roughness parameter pertaining to the Wenzel wetting state, is related to Sdr, the developed interfacial area ratios were very sensitive to microcavities, for their experiments using polypropylene textured using a femtosecond laser, the droplets were in neither Cassie-Baxter nor Wenzel modes.

Zhou et al. [55] explored laser remelting to improve the surface quality of LPBF-manufactured parts. In an AlSi10Mg sample, the initial sample displayed 70 deg contact angles; following laser remelting, the surface quality improved, and contact angle increased to 89 deg. They observed similar values of Sa, 17 µm on the top surface, comparable to the 15.52–21.71 µm observed in this study, although their side values of Sa (i.e., 6.5 µm) are lower than those observed herein. However, the authors did not report nor consider the impacts of build angle in their study, as the study focused on improving surface quality.

From Fig. 15, it may be seen that the Sdr decreases steadily with build angle until 30 deg. From 30 deg to 40 deg, the Sdr does not change significantly; however, the Sdr increases abruptly when going from a build angle of 40 deg to 50 deg. Then, from 50 deg to 60 deg, the Sdr decreases to a value less than that measured for the 40 deg angle. This trend suggests that there is a critical build angle between 40 deg and 50 deg with regards to Sdr. This spike in surface roughness at 40 deg–50 deg build orientations was demonstrated to a similar extent by Hovig et al. [57] for the LPBF of AlSi10Mg flat tensile specimens. They show a clear jump in surface roughness when going from a 30 deg build angle to a 45 deg build angle. Then, a decrease in surface roughness was observed for all angles greater than 60 deg. A similar spike in surface roughness was also observed, although to a lesser extent, by Karlsson et al. [58]. In their study, they investigated LPBF tool steel, and the surface roughness was shown to decrease rapidly with a build angle, but when going from a build angle of 40 deg to 45 deg, a flat trend resulted, while all other changes in build orientation provided for a marked decrease in roughness. Interestingly, Ullah et al. [59] found a steadily decreasing trend in surface roughness along the upskin of LPBF AlSi10Mg flat plates with the increasing build angles. Based on there being some inconsistency between various studies (including this one), it is not clear what the mechanism is for the sudden increase in surface roughness around the 50 deg build angle. Any criticality in surface roughness with respect to build angle would have to depend on the permeability of the powder bed and its resulting capillary force on the melt pool during contour track melting. With the help of gravity, the capillary force would have to be great enough to pull sufficiently against frictional and surface tension forces of the melt pool resulting in higher surface roughness due to partial wicking followed by solidification. There is a critical contact angle in which maximum wicking occurs for a fluid amongst a three-dimensional compact bed of spheres [60,61]. This critical contact angle is ~51 deg, and it is interesting to note that this is very close to the build angles in which surface roughness is maximum. This means that when/if the melt pool achieves a contact angle of ~51 deg, the capillary force becomes maximized. It is feasible to assume that molten AlSi10Mg can achieve such a contact angle while co-existing with its solid phase as similar contact angles have been reported in the literature [62]. In all cases, heterogeneities in the LPBF-manufactured components resulted in anisotropic wetting. Since the heterogeneities are defects of the manufacturing process, they do not exhibit uniform size nor spacing, thereby resulting in anisotropic wetting influenced by the part’s build angle. Vapor chamber designers should be aware of the anisotropy in LPBF-manufactured surfaces impacting wettability.

Water contact angles on AlSi10Mg surfaces manufactured via LPBF at various build orientations were investigated. Several qualitative and quantitative measurements have been provided. Although the powdered AlSi10Mg alloy is typically hydrophobic, wettability varied based on build angle: build angles of 0 deg and 60 deg exhibited primarily hydrophilic behavior, build angles of 10 deg and 20 deg demonstrated a mix of hydrophobic and hydrophilic behavior, and hydrophobic behavior was observed for 30 deg, 40 deg, and 50 deg build angles. The build/printing angle affects roughness of the surface as well as the size and the number of partially melted metal beads formed due to gravitational and powder-bed-capillary effects during the LPBF process. Surfaces printed at 30 deg and 40 deg printing angles had the smoothest overall surfaces and, therefore, the highest solid–liquid contact angles (i.e., consistently hydrophobic). Surfaces were not homogenous, and measured surface roughness, Ra, ranged from 5 to 36 µm, and arithmetic mean heights, Sa, ranged from 15.52 to 21.71 µm with variations based on location. The size of the partially melted metal beads decreased as the printing orientation angle increased. This information can be used to determine the desired printing angle of VCs. A significantly rougher surface was found for the 50 deg build angle. Based on the theoretical capillary behavior of a fluid with a compact bed of spheres, a critical contact angle of 51 deg results in the melt pool permeating the powder bed at a maximum. The critical melt pool contact angle and build angle in this case are near equal, suggesting a critical build angle exists, which gives rise to pronounced melt pool wetting behavior and increased surface roughness due to partial wicking followed by solidification. Based on this study, the effects of the printing angle did not conclusively change the surface roughness of different shaped geometries due to strong anisotropy in roughness observed. Due to the nature of LPBF, as well as the process factors used to produce the specimens in this study, the surface of the specimens contained significant mini-structures, which may prove valuable in the aid of heat transfer and bubble formation in VCs.

J. A. M., E. M. S.-C., and M. M. D. gratefully acknowledge the support of NSF (Grant No. 1828571).

There are no conflicts of interest.

The authors attest that all data for this study are included in the paper.