Additive Manufacturing of Cobalt-Based Dental Alloys: Analysis of Microstructure and Physicomechanical Properties

The limitations of investment casting of cobalt-based alloys are claimed to be less problematic with significant improvements in metal additive manufacturing by selective laser melting (SLM). Despite these advantages, the metallic devices are likely to display mechanical anisotropy in relation to build orientations, which could consequently affect their performance “in vivo.” In addition, there is inconclusive evidence concerning the requisite composition and postprocessing steps (e.g., heat treatment to relieve stress) that must be completed prior to using the devices. In the current paper, we evaluate the microstructure of ternary cobalt-chromium-molybdenum (Co-Cr-Mo) and cobalt-chromium-tungsten (Co-Cr-W) alloys built with direct metal printing and LaserCUSING SLM systems, respectively, at 0°, 30°, 60°, and 90° inclinations (Φ) in as-built (AB) and heat-treated (HT) conditions. The study also examines the tensile properties (Young’s modulus, E; yield strength, RP0.2; elongation at failure, At; and ultimate tensile strength, Rm), relative density (RD), and microhardness (HV5) and macrohardness (HV20) as relevant physicomechanical properties of the alloys. Data obtained indicate improved tensile properties and HV values after a short and cost-effective heat-treatment cycle of Co-Cr-Mo alloys; however, the process did not homogenize the microstructure of the alloy. Annealing heat treatment of Co-Cr-W led to significant isotropic characteristics with increased E and At (except for Φ = 90°) in contrast to decreased RP0.2, Rm, and HV values, compared to the AB form. Similarly, the interlaced weld-bead structures in AB Co-Cr-W were removed during heat treatment, which led to a complete recrystallization of the microstructure. Both alloys exhibited defect-free microstructures with RD exceeding 99.5%.


Introduction
e high strength and stainless nature of ternary cobaltchromium-molybdenum (Co-Cr-Mo) and cobalt-chromiumtungsten (Co-Cr-W) alloys, first patented in 1907 by Haynes, led to the development of compositional variants of cobaltbased alloys for clinical applications such as artificial knee and hip joints and denture frameworks for fixed and removable prostheses [1]. In dentistry, the first use of a cobalt-based alloy for investment casting is reported to have started in 1936 [2].
is was three decades after William Taggart invented a practical method for casting gold inlays [3]. Although casting or "lost-wax" technique is still popular in the dental technology, the steps involved are time-consuming and fraught with processing variables. In light of this, the advent of selective laser melting (SLM) could indeed be considered a watershed as it is claimed to offer significant benefits in terms of manufacturing speed. Since the devices are digitally built, they are likely to produce predictable clinical outcomes and be much closer to manufacturers' specification [4][5][6]. SLM is a metal additive-manufacturing (AM) process that employs high-energy laser beams to melt powders together, in a layer-by-layer fashion, into three-dimensional (3D) objects. However, it is a complex thermophysical process, which is dependent on many parameters including alloying constituents, machines, and the parameters of the controlled environment [6]. One of the potential limitations of the process is that the devices are likely to display mechanical anisotropy in relation to build orientations and this could affect their performance "in vivo" [7][8][9][10]. While a few studies [11,12] have attempted to investigate this limitation, there is inconclusive evidence concerning requisite composition and postprocessing steps (e.g., heat treatment to relieve residual stress) [13] that must be completed prior to using the devices. In the current paper, we evaluate the microstructure of commercially available ternary Co-Cr-Mo (ST2724G) and Co-Cr-W (Remanium CL) alloys, marketed for dental application. Samples were built with direct metal printing and LaserCUSING SLM systems, respectively, at 0°, 30°, 60°, and 90°inclinations to the layering (polar angle Φ) in "as-built" (AB) and "heat-treated" (HT) forms. e study also evaluates the tensile properties (Young's modulus, E; yield strength, R P0.2 ; elongation at failure, A t ; and ultimate tensile strength, R m ), relative density (RD), and microhardness (HV5) and macrohardness (HV20) as relevant physicomechanical properties of the alloys.

Specimen Preparation.
AB and HT Co-Cr-Mo samples (Co Bal Cr 29; Mo 5.5; Si < 1; Mn < 1; Fe < 1 wt.%) [14] were built with Type 5 ST2724G alloy (SINT-TECH Parc Européen d'Entreprises, Rue Richard Wagner 63200 Riom, France), whereas Co-Cr-W counterparts (Co Bal Cr 28; W 9; Si 1.5; Mn < 1; Fe < 1; Nb < 1; N < 1 wt.%) [15] were built with the Remanium star CL alloy (Dentaurum GmbH & Co. KG, Turnstraße 31, 75228 Ispringen, Germany) at 0°, 30°, 60°, and 90°inclinations, subsequently referred to as polar angle (Φ). e ST2724G raw powder had a powder density of 4.1 g/cm 3 and an average particle size (D 50 ) of 8-9 µm. Figure 1 shows the schematic of sample orientations. For Co-Cr-Mo specimens, a dumbbell-shaped CAD model (SolidWorks; Dassault Systèmes SolidWorks Corp., 300 Baker Avenue, Concord, MA 01742) designed in accordance with ISO 22674 requirements [16] was sent to 3D Systems (Rue Richard Wagner, 63200 Riom, France), and the samples were built at a layer thickness of 30 µm in an inert environment using a ProX DMP 200 printer [17]. e heat treatment for Co-Cr-Mo was completed in an airy furnace at 800°C for 20 minutes by the manufacturer (as recommended for dental devices) prior to the removal of the support structures (Mark Barnes Engineering, 35 Olympic Circuit, Southport QLD 4215, Australia) Unlike the directly built Co-Cr-Mo specimens, Co-Cr-W samples were supplied as 7 mm × 55 mm cylindrical bars built in an inert environment using an Mlab printer (Concept Laser GmbH, An der Zeil 8, 96215 Lichtenfels, Germany). Heat treatment was completed by the supplier in a furnace under argon atmosphere at a rate of 400°C/hour to 1150°C followed by a 1 h dwell time before cooling down to 300°C [18].
is particular treatment is recommended by Concept Laser for the Co-Cr-W alloy for dental applications. e cylindrical samples were turned on a CNC milling machine (Griffith Technical Solutions, Nathan Campus, Griffith University, 170 Kessels Road QLD 4111, Australia) using cooling emulsion to minimize possible alterations to their microstructure.

Microstructural Analysis.
e microstructure analysis involved optical examination of randomly selected test specimens for grain size, porosity and voids, dendritic growth, cracks, and other defects. Four grinding and mechanical polishing steps were performed to expose the metallurgical structure. e steps comprised grinding with silicon carbide (SiC) 220 sandpaper, polishing process with 9 µm and 3 µm diamond suspension, and surface finishing with a silicon monoxide (SiO) solution with a grain size of 0.5 µm. e visibility of the scan track pattern and the inherent grain structure was enhanced by a subsequent etching process. Two etching procedures were employed: etching with "Beraha 3" acid (comprising 60 ml H 2 0, 40 ml hydrochloric acid, 5 g ammonium hydrogen difluoride, and 1 g potassium bisulphite) and electrochemical etching at 3 V for 10 seconds in a 3% hydrochloric acid solution. Highresolution images of the etched microsections were taken with an optical light microscope (Axio Imager z2m; Carl Zeiss Microscopy GmbH, Jena, Germany). In determining the RD, area porosities were calculated on the micrographs with the Zeiss AxioVison software package (Carl Zeiss Microscopy GmbH, Jena, Germany).

Tensile Test.
Tensile testing to assess Young's modulus (E), 0.2% yield strength (R P0.2 ), elongation at failure (A t ), and ultimate tensile strength (R m ) was performed as per ISO 22674:2016 guidelines [19] Test specimens (n ≥ 3) from each configuration and condition (inclination and treatment)  were subjected to uniaxial tensile loading and stressed until failure occurred.. Specimens (gauge diameter 3 mm, gauge length 15 mm, and radial shoulders) were tested to yield at a maximum load of 10 kN and 1.

Results
For microstructure analysis, only images from crosssectional micrographs are presented in the paper due to the configurational similarities and corresponding redundancy of longitudinal micrographs ( Figure 2). For instance, the cross-sectional cut of a 0°orientated specimen is similar to the longitudinal cut of a 90°specimen; likewise, the cross-sectional cut of a 30°inclined specimen matches the longitudinal cut of a 60°specimen. Figure 3 shows the obtained area porosities of both alloys, measured on the microsections. e relative density (RD) is 99.93% for Co-Cr-Mo and 99.86% for Co-Cr-W, respectively. Relative densities obtained via microsections result generally in higher percentages compared to the determination via the Archimedes method [22]. On the 90°inclined Co-Cr-W sample, small delaminations induced spikes in area porosity. In Figures 4 and 5, the microstructures of Co-Cr-Mo and Co-Cr-W specimens are depicted for each inclination and both conditions, i.e., AB and HT conditions for the left and right columns, respectively. e tensile test data are summarized in Table 1, while Figure 6 shows representative stress-strain graphs of the alloys. Figure 7 compares and summarizes Young's modulus (E) data of the alloys after heat treatment as per dependency on the inclination angle. Figure 8 presents the graphical summaries of the yield strength (R P0.2 ), elongation at failure (A t ), and ultimate tensile strength (R m ). In addition, the corresponding minimum requirements stated in the ISO 22674 standard [16] are included in the graphical summaries as well. Lastly, microhardness (HV5) and macrohardness (HV20) results are summarized in Figure 9.

Discussion
While in the mouth, dental devices are subjected to stress from masticatory actions; hence, alloys should show high recovery capacity, resilience, high specific stiffness (E/RD), and elongation, if devices are to withstand biting forces without fracture [23]. Figure 10 shows an illustration of a denture framework with different functional parts, such as clasp components and connectors. ese parts explain the importance of orientation-dependent fluctuations in the material characteristics, which are present in selective lasermelted metals. e standards for assessing the mechanical properties of dental alloys [19] place emphasis on E, A t , and R P0.2 of the alloys. In a clinical context, R P0.2 is relevant for predicting failure in multiple unit dental bridges. A t determines how the alloys can sustain a large permanent deformation before they fracture. It is worth noting that A t could serve as a quality control measure that verifies the level of impurities and proper processing of the devices, while E measures the alloy's ability to resist flexure, especially in metalceramic restorations where flexure will cause fracture of the porcelain [24]. Literature about alloys provides information on their hardness; while a low value cannot resist occlusal forces, an extremely high value could make the alloy difficult to grind and polish and even wear the Cross-sectional Advances in Materials Science and Engineering opposing teeth [23]. Table 2 summarizes the tensile properties as a range, i.e., respective minima and maxima for alloys examined in comparison to data from alloy manufacturers and standard requirements for dental alloys.

Microstructural
Analysis. SLM-produced Co-Cr alloys are reported to display superior homogeneity compared to cast Co-Cr alloys, which are inherently associated with large initial grain size, nonhomogeneities, and other casting defects [6]. e evaluated area porosities on the micrographs revealed average RDs greater than 99.8% and individual RDs greater than 99.5% in all cases. Since the applied heat treatment in this study did not comprise a hot isostatic pressing (HIP), which could enhance the RD further, it can be concluded that the post-heat-treatment did not alter the RD of the specimens [26]. It is worth stating that the observed density does not negate a possible inhomogeneity at the increasing build height [9]. Given that the criterion applied in the aeronautics industry for

Advances in Materials Science and Engineering
RD is greater than 99%, these values could be considered to be exceptional for the manufacturing process [27]. For most applications, such a high RD is not required and a correlated reduction in mechanical strength is traded for speed and economic fabrication [28][29][30]. However, for dental metallic devices, the RD is also relevant to accomplish a tissue-friendly, void-free satin finish, which is achieved via electrolytic polishing to prevent fitting and cleaning problems on the fitting surface of denture frameworks. Given that in SLM the residual porosity tends to be common in the subsurface, which is the transition area between the contour and core irradiation, lower RD could potentially lead to laid open pores after polishing [10, 31, 32].

As-Built Condition.
In the AB condition, the stacking scheme of the individual weld beads as well as the applied irradiation strategy controls the micrographs (left-hand side in Figures 4 and 5). e absence of needle-like-appearing fine-structured platelets, seen in cast samples, is noteworthy. ese are formed at thermodynamic equilibrium conditions at 417°C alongside the allotropic transformation from the face-centered cubic (fcc; c-Co) to the hexagonal close packed (hcp; ε-Co) crystal structure [33]. In investment casting, cooling rates are very slow within the pool of molten ingots, and larger grains are formed around tiny nuclei that grow until the grain boundaries meet the solid state. Because of this, dendritic structures may be very large in cast Co-Cr alloys, and the size of a single grain can approach the  diameter of a removable partial denture framework clasp [5]. Due to the rapid cooling rates in SLM, no large grains or dendritic structures are evident. Moreover, the absence of the platelets indicates that the fcc to hcp transformation did not take place, and the thermodynamic equilibrium was not reached. is suggests that the metastable c-Co is present in the as-built condition, which is comparable with a recently published study on heat treatments of additively manufactured Co-Cr-Mo. According to Kajima et al. [34], ε-Co is only present after an additional heat treatment and the volume fraction between c-Co and ε-Co varies upon the heattreatment temperature, with c-Co remaining as the predominant phase (≥75% volume fraction). e transformation of c-Co to ε-Co is dependent on the grain size, and for finegrained structures (as is the case in SLM), this transformation is suppressed, resulting in the predominant c-Co [35].

Heat-Treated Condition.
Heat treatments and modification of elemental composition could change the microstructure and consequently alter the physical-mechanicalchemical properties of Co-based alloys. In SLM, heat treatment is required to remove accumulated thermal stresses in long-span devices [13]. It is worth clarifying that the cycles for both alloys differ significantly. Co-Cr-Mo was heat treated at 800°C, which induced slow-progressing recrystallization, whereas at 1150°C for Co-Cr-W, a soft annealing with directional recrystallization occurred. e latter led to a complete restructuration of the microstructure, whereby the stacking scheme of single weld beads was no more visible ( Figure 5). At 1150°C, Cr and W are solved in c-Co and a chemically homogeneous situation is given. is step was followed by a very slow cooling (furnace cooling), which ensured thermodynamic equilibrium conditions. In general,

Advances in Materials Science and Engineering
the Co-Cr-W material displayed predominantly isotropic properties with only minor fluctuations after heat treatment, while E was raised significantly and strength and hardness were reduced. Savage [36] explained that the preferred temperature range for the isothermal transformation (notably by lamellar constituent) of the ternary Co-based alloys is about 1040°C to 1100°C and the length of the isothermal treatment is a function of composition of alloy and temperature of transformation, but in most instances, a one-hour dwell time at the target temperature is sufficient. e isothermal heat treatment at 800°C for 20 minutes of Co-Cr-Mo resulted in significantly improved E and slightly increased R P0.2 , A t , R m , and HV. An assumption of the origin of the higher and unsystematically varied R m was given in a study by Turrubiates-Estrada et al. [37]. It was indicated that heat treatment at 800°C for 20 minutes promotes the formation of ε-Co and intermetallic precipitations and/or carbides ( Figure 11, [38]). According to Kajima et al. [34], heat treatment at 800°C for 6 h can lead to a volume fraction of around 20-25% ε-Co in SLM samples. e formation of the second phases in a sufficient fine distribution (high undercooling) could be the origin of the observed increase in the mechanical values. It should be noted that, during the additive formation, the material undergoes several heating cycles. After the main fabrication step, i.e., the raw metal powder being completely molten and rapidly solidified, the solidified volume passes a continuous time-temperature-transformation (TTT) as shown in Figure 12, [38]. During cooling, the observed volume undergoes several postheating cycles by additive formation of further layers on top of the observed volume; hence, it is possible that a considerable fraction of ε-Co can be generated via alterations in the process conduct [10].

Tensile Properties.
All test specimens yielded at the ultimate stress without localized necking. Specimen fracture occurred along the gauge length, indicating the homogeneous nature of the alloys. e overall polar angle dependence of the mechanical properties was low, except for A t , which showed large fluctuations in the AB condition for both alloys and for Co-Cr-Mo in the heat-treatment condition. ese findings confirm the assertion that ductility represents the most volatile tensile characteristic. Our data for polar dependency at Φ � 90°corroborate findings by Takaichi et al. [12], but contrast studies that examine Albased [9] and Fe-based [8] metals. is shows that in SLM characteristics on the anisotropy vary and are dependent on raw materials.
Although there are no existing manufacturer data on build orientations for the Co-Cr-Mo alloy used in our study,   Advances in Materials Science and Engineering our data show that Co-Cr-Mo in the AB form required heat treatment to attain the minimum E value as per standards requirements (except for Φ = 60°). Despite the improved tensile properties and HV values, the short and costeffective heat-treatment cycle did not completely homogenize the properties of alloys. Both alloys showed increased E after heat treatment, which is related to the increased ratio of ε-Co. Differences in E values for the two alloys are probably due to varying ratios of c-Co to ε-Co in the respective structure. Since the phase content is not absolute, it may change during the tensile test via straininduced phase transformation, and the ε-Co fraction can increase [12]. e precise influence of this effect needs to be studied further. Manufacturer-provided density at all orientations is 8.6 g/cm 3 ISO 22674:2016 [16] 150 500 2 * * * * Not provided Advances in Materials Science and Engineering e divergences observed in AB conditions are possibly due to varying irradiation in the printing or manufacturing systems. In the heat-treated condition, the differences observed are related to the partial recrystallization that occurred at 800°C in comparison to the full recrystallization at 1150°C followed by furnace cooling, allowing thermodynamic equilibrium conditions. Apart from the HV values, E, R P0.2 , A t , and R m are higher in Co-Cr-W than in Co-Cr-Mo.
is simultaneous increase in their respective AB conditions could be an indication that the C-formed curves of the TTT diagram ( Figure 11) are moved to lower times for the Co-Cr-W alloy. Applying the annealing heat treatment, E and A t (except for Φ = 90°) increased, whilst R P0.2 , R m , and HV of Co-Cr-W decreased.
Concerning the standards requirements for Type 5 dental alloys, both alloys can be considered suitable for clinical use in their heat-treated condition. Although AB Co-Cr-W also fulfills these requirements, a heat treatment is highly recommended, not only to minimize possible distortions from residual stress, but also to enhance the ductility and stiffness of the material. To guarantee the integrity of the built parts, it is recommended that heat treatment be performed on built parts whilst they are firmly attached to the support structure on the substrate plate. E and R m values provided by the Co-Cr-W manufacturer are somewhat comparable to ours. A t values at 0°and 90°in this study are higher than those of the manufacturer, whereas our R P0.2 values are lower. e higher and arbitrary R m values for the Co-Cr-Mo alloy may be due to the increased surface roughness of the specimens or the possible subsurface porosity acting as crack initiation [9]. However, these prospective "inhibitors" were absent in the machined Co-Cr-W specimens with documented studies suggesting the machining process results in improved mechanical properties by enhancing R m by 5-10% and E up to 10% [39][40][41].

Hardness.
In our study, we observed minor deviations between microhardness (HV5) and macrohardness (HV20). However, data from the microhardness (HV5) tests might be prone to fluctuations due to limited indentation depth (approximately 20 μm at 380 HV5), which is lower than a single layer (∼30 μm) and thus may not be a conclusive measure ( Figure 13). For comparison, the indentation depth for 380 HV20 is approximately 40 μm. e hardness tests did not reveal any significant fluctuations regarding the sample orientation. In addition, both alloys exhibited similar HVs in their AB conditions, despite the differences in their R P0.2 and R m values. In their respective HT conditions, HV values differ (∼435 HV for Co-Cr-Mo and ∼360 HV for Co-Cr-W), but with comparable R m values. e disparity between HV values and tensile properties is possibly because hardness testing focuses on compressive stresses, whereas the tensile test produces failure at uniaxial forces. It is documented that the mechanical properties of SLM-produced alloys do differ in regard to tensile versus compressive loading [42]. Moreover, it is known that the conclusion from hardness to tensile strength is problematic for selective laser-melted metals [10].

Conclusion
In this study, we assessed the clinical properties of lasermelted Co-Cr-Mo and Co-Cr-W alloys in as-built and heattreated conditions. e anisotropy observed in the alloys primarily influenced ductility in addition to minor fluctuations in tensile strength and Young's modulus. In the as-built condition, the alloys displayed reduced Young's modulus and hence required heat treatment to meet clinical requirements set out in the standards. Similarly, heat treatment is recommended if the desired ductility and isotropic characteristics are to be guaranteed. Although the study confirms the capacity of the SLM systems to produce consistent parts under controlled manufacturing parameters, further

Data Availability
Data obtained for this work are provided in the results section. When additional data/results were taken to validate or confirm statements, the source has been cited accordingly.

Disclosure
Leonhard Hitzler and Frank Alifui-Segbaya are co-first authors. Advances in Materials Science and Engineering 11