Metallurgical Characterization of Co-Cr-Mo Parts Processed by a Hybrid Manufacturing Technology

The additive manufacturing technology offers new and incredible opportunities in the design of components. Nowadays, structural integrity assessment of additively manufactured components is a formidable challenge that needs to be faced out in order to allow such components to be launched in the market. One of the major drawbacks of additive manufacturing is poor surface finish and loose geometrical tolerance of built parts. In this scenario, hybrid manufacturing, which takes advantage of both subtractive and additive manufacturing, can be considered as a solution worthy of investigation in view of possible applications to save costs and time in the component production. The present work is aimed at assessing microstructural properties of Co-Cr-Mo specimens manufactured by the hybrid subtractive/additive technology, when the additive part is built over the machined one. The results show an excellent metallurgical coupling at the interface between the two differently processed parts.


INTRODUCTION
In recent years, additive manufacturing (AM) technologies applied to metallic materials have experienced a formidable breakthrough both in the industrial sector with the production of complex structural parts and in the scientific as well as academic sector where the challenges were, and still are, to find new suitable-to-process alloys [1][2][3], develop process parameter optimization criteria [4,5], assure structural integrity of parts [6][7][8][9][10][11], and redesign components with the aim of reducing their weights. These technologies, say, selective laser melting (SLM), electron beam melting (EBM) or binder jetting (BJ), offer numerous advantages over standard subtractive methodologies: (i) a complete freedom in component shape design, (ii) less production of scraps, (iii) shorter time to market, and (iv) more customization opportunities. Considering the possibility of reducing the weight of parts and saving materials, AM can be classified as a green technology as well. However, there are disadvantages, the most important of which are high residual stress, poor dimensional accuracy and surface finish, high initial costs, and long time for process parameter optimization. Low productivity could also be a drawback when considering the SLM process [12]. In fact, using optimum process parameters and the layer thickness 40 μm, the volumetric deposition rate of maraging steel during SLM with an EOS M290 machine is 10.8 cm 3 /h [13,14].
In this regard, conventional subtractive manufacturing technologies are irreplaceable and even strategic when used in the synergic combination with additive techniques. Hybrid additive/subtractive manufacturing (A/SM) of metallic parts was introduced in 1998 with the pioneering work of Kruth et al. [15], who used CNC (computer numerical control) milling to remedy the relative inaccuracy of each powder jet deposition during component building. It should be noted that, when dealing with hybrid manufacturing, subtractive techniques can be used either in parallel with additive techniques or in series. In the first case, the tolerance and roughness of a part is adjusted, for example, by milling during building the whole component [16][17][18]. In the second case, a component portion that needs high dimensional tolerance is first produced, say, by turning, above which the remaining part of the component that does not require high surface finishing is built by additive techniques (this case is studied in the present work). The procedure in series is relatively recent in the panorama of the powder bed fusion (PBF) technologies, and it is obviously much different from the standard use of finishing processes applied to the already built additive part. Among the main advantages of hybrid systems are the repair of the currently existing high-value components, high surface finish of external and internal parts of complex shape, the precision guaranteed by the possibility of printing and milling a part in the same reference coordinate system, improved fatigue strength due to high surface finishing of components [19], improved productivity, and finally multimaterial 3D printing [20]. For a short review of hybrid additive systems, the readers can refer to [21].
Marklein et al. [37], in their recent work, described the manufacturing of a gear component with discrete tooth geometry by powder bed fusion of metal with a laser beam and forming. They found that the tooth geometry formed by additive manufacturing and forming is closer to the target than for the conventional process chain. Dolev et al. [38] examined in detail the tensile behavior and fracture toughness of a hybrid Ti-6Al-4V alloy. Compact tension and uniaxial tension specimens extracted from the hybrid preforms demonstrated good fracture and properties with no preference for crack growth in AM or wrought materials. The impact of process parameters on the relative density, warpage, and bonding strengths between the sheet metal and additively manufactured element was studied in work [39]. Meiners et al. [40] investigated the use of laser metal powder deposition (LMPD) and wire-arc additive manufacturing (WAAM) for the hybrid production of aerospace Ti-6Al-4V forgings. The proposed manufacturing route is based on a conventionally preformed forging, which does not yet have all the features of the final component. These features, such as ribs or other structural or functional geometries, will be added by additive manufacturing. In this way, the number of processing steps and forging dies are reduced, and the efficient near-net-shape production could be provided.
A/SM parts show high mechanical properties as compared to both simple AM specimens and wrought alloy. For instance, Du et al. [16] analyzed the mechanical and metallurgical properties of an 18Ni maraging steel mould produced by a novel method of additive and subtractive manufacturing of metallic parts (SLM and milling occurred alternatively). Using this technique, internal cooling water channels were formed and machined as well. The measured hardness values were much higher than those of the just SLM and wrought maraging steels.
Tan et al. [41] carried out metallurgical and mechanical characterization of a bimaterial specimen made of SLM-processed maraging steel above the machined copper part. They also focused on the interface bonding mechanism. The good metallurgical bonding was promoted by the Fe and Cu interdiffusion. Gradient sub-micrograins with the 111 texture at the interface were detected by EBSD, and finally tensile fracture occurred in the wrought copper part, far from the interface, demonstrating excellent properties of the functional bi-material component. Osipovich et al. investigated the microstructure formation in a wire-feed electron-beam additively manufactured bimetallic steel-copper specimen designed to obtain a transition zone between the two difficult-toweld alloys [42]. Azizi et al. studied a possibility to deposit the maraging steel powder onto C300 maraging steel as well as the H13 tool steel substrates [43]. In the first case, the deformation was homogeneous until failure localized away from the interface; in the second case, the deformation was predominantly within the substrate until failure localized at the interface.
The possibility to combine AM with sheet metal forming for the productivity improvement was also explored in [44][45][46][47]. Potential applications may be found in the production of individualized functional elements on the formed sheet in the field of medical and aerospace industry.
Nowadays, AM technologies are used to fabricate biomedical parts thanks to their unique advantages particularly suitable for such kind of products, like customization. A comprehensive review of this topic was recently proposed by Revilla-León et al. [48]. They focused on the advantages of AM techniques compared to the traditional ones, such as casting and subtractive methods, for the production of dental implants. However, as pointed out by the same authors in the Conclusions section of their paper, "additional clinical studies are required to assess the long-term clinical performance, biological and mechanical complications, and prosthetic restoration capabilities of additively manufactured dental implants. Moreover, further studies are needed to evaluate their long-term success and survival rates and biological and mechanical complications of AM implantsupported prostheses". The same conclusions were reached by Mangano et al. [49] who studied the suitability of direct metal laser sintering to the production of titanium dental implants. Silva et al. [50] demonstrated the possibility to use the hybrid additive manufacturing process composed of SLM and stereolithography (SLA) to produce customized dental implants. In particular, the teeth core was replaced by a topologically optimized metallic mesh produced by SLM in order to obtain the desired mechanical strength of the part. The metallic mesh was then covered by a polymeric material using SLA, which reproduces the teeth crown based on the anthropometric data of the patient. A novel and original implant design, not yet explored in this scenario, is based on the use of a hybrid turning/SLM process to produce dental implants. A hybrid implant abutment is shown in Fig. 1, where the implant coupling interface (i.e. T-base), with its stringent dimensional and geometric accuracy and surface roughness requirements, is manufactured by a standardized and highly productive turning process while the upper part, where the substructure is intended for veneering and which should mimic the occlusal morphology of patient's dentition, is additively deposited by SLM. In these applications, the interface between the two parts should be well characterized.
This work is therefore aimed at characterizing, from a metallurgical and mechanical point of view, the interface between the turned cast and additively manufactured parts. The material investigated is Co-Cr-Mo alloy, which is typically used as the substrate material for dental crowns. Xiang et al. [51] found that the microstructure promoted by selective electron beam melting (SEBM) consists of face-centered cubic columnar grains with the 001 preferred orientation and continuous thin carbide films at the grain boundaries. This induces a strong anisotropy of mechanical properties and suggests that the SEBM-built Co-Cr-Mo along the build direction is suitable for load-bearing orthopaedic implants.
Results obtained in the present work show the epitaxial grain growth at the interface of the two differently processed parts with an excellent metallurgical bonding at the interface.

Specimen Production
Hybrid cylindrical specimens (Fig. 2) were produced in two steps. The first half of a hybrid specimen was obtained by turning a cylindrical cast bar. The second half was produced by consolidating different powder beds over the first one using a laser  source (selective laser melting technique), whose parameters are summarized in Table 1.
Cylindrical specimens were used exclusively for metallurgical investigations through an optical microscope.
The chemical composition of the two halves of cylindrical specimens is collected in Table 2. ASTM F75 belongs to the family of cobalt-based alloys that have been widely used as materials for valve seats in nuclear power plants, aerospace fuel nozzles, and engine vanes as well as orthopaedic and dental implant materials, because of their strength at high temperature, corrosion resistance, excellent wear resistance and biocompatibility. An extensive characterization of this alloy fabricated by electron beam melting was performed by Sun et al. [52,53].

Experimental Procedure
Specimen roughness was measured according to Standards ISO 4287:1997 and ISO 4288:1998. In particular, the measurement was carried out considering only the middle part of the specimens with the total evaluation length 21 mm (10.5 mm each half). In order to increase the statistical significance, each specimen was measured three times, in three different positions, rotating the specimen by an angle of 120° around the longitudinal axis each time.
Metallurgical investigations were carried out on the longitudinal cross section of the specimen. The analysis was focused above all on the interface between the subtractively and additively manufactured parts. Optical (Leica LM2500) and scanning electron microscopes (SEM, a Quanta FEG-250 model of FEI energy dispersion spectroscopy (EDS, EDAX ©) and electron backscattered diffraction (EBSD, EDAX©)) were used for microstructural characterization. Before investigations, the specimens were hot embedded in phenolic resin and prepared according to the standard metallographic procedure up to final polishing with the colloidal silicon oxide suspension. Chemical EDS analyses were carried out with the acceleration voltage of the electron beam 20 kV; EBSD scanning was performed over the scan area ~500 × 1500 µm 2 with the step 2 µm. For optical microscope observations, the microstructure was highlighted by electrochemical etching with the water solution of hydrochloric acid and continuous current.

RESULTS AND DISCUSSION
Optical micrographs (Fig. 3) highlight the microstructural morphology of the two differently manufactured parts. The typical pool shape of each track promoted by SLM as well as dendritic grains coming from the alloy solidification process inside the mold are observed. Figure 3 also shows the interface between the two parts where no discontinuities were detected.
The SEM micrograph in Fig. 4a shows the specimen macrostructure in the longitudinal cross section. Several interconnected porosities with irregular morphology are observed especially along the central axis of the as-cast alloy. From the morphological analysis, it is evident that such porosities are due to the lack of filling of interdendritic areas by the liquid metal in the solidification phase (shrinkage porosity). In addition, in the as-cast material, it is easy to observe the growth direction of columnar grains, which  is parallel to the heat dissipation direction by the mold (Fig. 4b). In fact, as also supported by fractography of the as-cast material (Fig. 5), columnar dendritic grains develop radially from the surface towards the center of the specimen. In particular, at a distance of about 500 µm from the center (Fig. 5a), visible interdendritic porosities are shown (Fig. 5b) due to the lack of the liquid metal supply at the last solidification phase. In areas closest to the mold wall, where solidification took place first, no porosity was observed (Fig. 5c). The radial growth of columnar grains in the as-cast material explains why, in the longitudinal section of the specimen, the columnar grains assume an increasingly equiaxed morphology towards the center of the specimen (Fig. 4c). In the material produced by SLM, isolated porosities were observed, mainly spherical and with a dimension lower than 30 µm. At higher magnification  ( Fig. 4d), one can see shapes of molten pools and columnar grains inside them, which developed along the heat dissipation direction. At the as-cast/SLM interface, columnar grains of the selective laser melted alloy are predominantly oriented perpendicular to the interface; moreover, there does not seem to be an evident microstructural discontinuity at the interface (Fig. 4c). For a more in-depth investigation of the transition zone between the two parts, EDS mapping with chemical element distribution as well as line scan analysis were carried out. The results are shown in Figs. 6 and 7, respectively.
As reported in [54,55], the Co-Cr-Mo alloy is constituted by the CFC α-Co matrix and interdendritic or grain boundary carbides (M 23 C 6 type). EDS maps in Fig. 6 show how, in the as-cast material, the interdendritic phase is richer in Mo and Cr as compared to the α-Co matrix. Due to the low C content in the alloy, which is difficult to detect with the EDS technique, it was not possible to highlight any chemical variations between the dendritic and interdendritic regions. However, the presence of a higher Mo and Cr content in the interdendritic regions suggests the precipitation of the (Cr, Mo) 23 C 6 phase. This hypothesis is supported by Tan et al. [56] who detected columnar carbide chains or clusters consisting of nanosized cuboidal particles, which precipitated coherently with the surrounding γCo phase within interdendritic regions. By comparing the element distributions of the two materials, it turns out that the powders used for SLM are richer in W and Si than the as-cast alloy, which in its turn shows a higher content of Cr, Fe, and Ni. However, due to the element diffusion induced by high temperatures reached during different laser passes, W, Si and Cr, Fe, Ni show a concentration gradient between the as-cast and SLM materials, which increases for W and Si and of the opposite trend for the other 3 elements, as indicated in the concentration profiles in Fig. 7. Since no evident variation of the Mo and Co concentrations was noticed, these elements seem to show a more contained, or even absent, diffusive phenomenon; this is probably due to similar Mo and Co concentrations in the two different raw materials. From the EDS results, it can therefore be observed that there is no clear chemical discontinuity at the interface between the two specimen parts.
In the SEM micrograph of Fig. 6, it is also possible to notice the presence of some cracks in the direction perpendicular to the interface. It is likely that rapid melting and solidification cycles of the first powders deposited on the as-cast material promoted high thermal stresses, therefore inducing hot solidification cracks. In this regard, it should be pointed out that such defects can be avoided by means of an adjustment of process parameters (say, an optimization of volumetric energy density [4]) or eventually by means of subsequent hot isostatic pressing (HIP) [57,58]. More in detail, in order to reduce the highest residual stresses at the beginning of the powder deposition and therefore the risk of hot tearing, the turned part surface could be preheated by laser scanning without powder. Further studies will be carried out in facing such issue in a next work. The high concentration of O and Si in one of the cracks should be attributed to the incomplete removal of the colloidal silica suspension used for the specimen polishing. The EBSD analysis carried out at the interface between the two different parts allows comparing the orientation of crystalline grains at different distances from the specimen axis. In addition to crystallographic orientation maps, polar figures (PF) and inverse polar figures (IPF) for the two differently processed parts are reported in Figs. 8 and 9. It is worth mentioning that distributions of crystallographic orientations in the polar figures (Figs. 9c-c″ and 9d-d″) are affected by the rotation by about 10° around di-rection [001] orthogonal to the specimen. The EBSD mapping was in fact performed for a specimen not perfectly aligned to the scan (Fig. 10), as shown by the slight inclination of columnar grains near the specimen boundary. However, the rotation around direction [001] orthogonal to the scan area has no influence on IPFs.
From the polar figures (PFs) of the as-cast material (Figs. 9c-9c″), it is observed that columnar grains show a very evident preferential orientation. It is known in fact that, during solidification, the columnar grains with a cubic lattice generally grow along direction 001   [59]. In the as-cast material, the texture of the columnar grains in the center of the specimen with the growth direction (GD) perpendicular to the section is in fact expected (Fig. 7c). Near the specimen surface, the expected growth direction of the columnar grains was 001 ,   which is parallel to the micrograph. However, the relative po-  Fig. 9c″ shows a texture that deviates slightly from the expected one (Fig. 11); such deviation is both due to the already mentioned inclination of the specimen and due to imprecise longitudinal sectioning of the specimen, which didn't take place on a plane containing the specimen longitudinal axis (Fig. 12). It is therefore evident that this plane does not cut the columnar grains along the {100} planes as one of the possible planes of symmetry.
Comparing the PFs (Figs. 9c-9c″ and 9d-9d″) and IPFs (Figs. 9e-9e″ and 9f-9f″) obtained from the two differently processed materials, it is noted how the columnar grain texture in the material produced by SLM agrees with that of the as-cast material. This is particularly evident in the central area and halfway between the specimen core and its boundary. The maximum intensity values of the PFs and IPFs coming from the two parts of the specimens are completely comparable as well.
The orientation relationship near the specimen surface is not so evident as in the two previous posi-    tions. However, the polar figures of the two differently processed parts ( Fig. 9c″ and 9d″) show a maximum intensity at about 73° from direction [001]. Moreover, the other two maxima in the PF of the SLM material are aligned with the band observed in the PF of the as-cast material. In general, it can therefore be stated that the grains formed by solidification of the fused powders adjacent to the as-cast material grow in an epitaxial way [60] because their orientation agrees with that of the grains of the underlying material.
The roughness of additive and subtractive parts was measured to be 3.925 ± 0.860 and 0.855 ± 0.416 μm, respectively, with the cut-off length 0.8 mm. The obtained great difference expectedly results from the different process technology used with the major detrimental effect on the roughness coming from SLM compared to the turning process.

CONCLUSIONS
The metallurgical assessment of hybrid specimens made of Co-Cr-Mo alloy first by turning a cast cylinder and then by adding the alloy on the turned part by the selective laser melting technology was carried out. The microstructure of the turned part is typical of components produced by casting, with dendritic grains growing in the direction of heat dissipation. Moreover, they grew along direction 001 ,   which is typical of cubic crystal lattices. The microstructure derived from solidification of the first powder beds was influenced by grains of the turned part at the interface, thus giving rise to an epitaxial grain grow, like in welding solidification phenomena. This ensured a perfect bonding between the two differently processed parts. Finally, to avoid hot tearing near the interface, an AM process tune-up will be necessary to define the process parameter transition from the cast to additive portion. Such issue will be addressed in a future work.

ACKNOWLEDGMENTS
We are grateful to Andrea Sandi (3D Fast Srl) for the specimen production as well as to Eng. Filippo Da Rin Betta and Eng. Federico Uriati who carried out roughness and tensile tests.

CONFLICTS OF INTEREST
The authors declare no conflicts of interest.