**Influence of Heat Treatments on Microstructure and Mechanical Properties of Ti–26Nb Alloy Elaborated In Situ by Laser Additive Manufacturing with Ti and Nb Mixed Powder**

#### **Jing Wei 1, Hongji Sun 1, Dechuang Zhang 2,\*, Lunjun Gong 2, Jianguo Lin 1,\* and Cuie Wen <sup>3</sup>**


Received: 25 November 2018; Accepted: 18 December 2018; Published: 25 December 2018 -

**Abstract:** In the present work, a Ti–26Nb alloy was elaborated in situ by laser additive manufacturing (LAM) with Ti and Nb mixed powders. The alloys were annealed at temperatures ranging from 650 ◦C to 925 ◦C, and the effects of the annealing temperature on the microstructure and mechanical properties were investigated. It has been found that the microstructure of the as-deposited alloy obtained in the present conditions is characterized by columnar prior β grains with a relatively strong <001> fiber texture in the build direction. The as-deposited alloy exhibits extremely high strength, and its ultimate tensile strength and yield strength are about 799 MPa and 768 MPa, respectively. The annealing temperature has significant effects on the microstructure and mechanical properties of the alloys. Annealing treatment can promote the dissolution of unmelted Nb particles and eliminate the micro-segregation of Nb at the elliptical-shaped grain boundaries, while increasing the grain size of the alloy. With an increase in annealing temperature, the strength of the alloy decreases but the ductility increases. The alloy annealed at 850 ◦C exhibits a balance of strength and ductility.

**Keywords:** heat treatment; in situ alloying; laser additive manufacturing; mechanical properties; microstructure; Ti–Nb alloy

#### **1. Introduction**

Beta (β) titanium (Ti) alloys containing non-cytotoxic elements such as niobium (Nb), zirconium (Zr), tantalum (Ta), tin (Sn), and molybdenum (Mo) have been intensively studied for applications in biomedical domains [1–5]. Among these, Ti–Nb alloys have attracted particular attention due to their high strength (600 MPa) and their very low elastic modulus (50 GPa), close to that of cortical bone (30 GPa), which was observed to reduce the stress-shielding phenomenon [6]. These alloys are considered to be potential substitute materials for conventional materials such as Ti–Ni or Ti–6Al–4V in order to prevent the release of toxic nickel (Ni), aluminium (Al), or vanadium (V) ions into the human body [7–11].

Recently, laser additive manufacturing (LAM), based on coaxed powder melting and rapid solidification through layer-upon-layer deposition, has attracted a great deal of attention in the fabrication of fully dense near net-shaped metallic components [12–15]. It has several significant advantages over traditional manufacturing methods, such as a wide range of material forms,

a simple manufacturing process for complex parts, short production cycle, raw material saving, excellent performance, etc. So it is an effective method for the personal customization of biological implants.

Generally, the powders for LAM are obtained from a prealloyed material that is converted into powder by plasma arc or gas atomization to achieve spherical particles with a particle size between 10 and 110 μm. In the case of Ti–Nb alloys, their high melting temperatures make it very complicated to produce powders in this way. So the Ti–Nb alloy powders that can be applied to direct laser metal-forming are still rare or even commercially unavailable. Recently, some researchers attempted to fabricate Ti–Nb alloys using selective laser melting (SLM) with mixtures of elemental powders of pure Ti and Nb. For example, Fischer et al. [16] fabricated Ti–26Nb (at.%) using SLM with elemental Ti and Nb mixed powders with high energy levels, and the alloy exhibited a homogeneous and dense microstructure with a β structure. Wang et al. [17] also fabricated Ti–Nb alloys with different Nb content using the same method and investigated the effects of Nb content on the microstructure and mechanical properties; they found that SLM could be used for in situ fabrication of Ti–Nb bone implants with tailored mechanical and biomedical properties by adjusting the level of Nb.

It has been reported that, compared to the microstructure achieved with conventional casting and forging, Ti alloys prepared using LAM usually exhibit a quite different microstructure due to repeated rapid solidification and rapid annealing during laser forming. Moreover, defects can be found in the as-deposited part, such as inconsistency in the structure and stability of mechanical properties, residual stress, and pores, which can weaken the mechanical properties of the alloys. Post-heat treatment can have an important influence on the microstructure and properties of the alloys [18–20]. However, microstructure transformation after heat treatment of as-deposited Ti–Nb alloys has rarely been investigated. Therefore, a deeper understanding of the microstructural evolution of Ti–Nb formed by LAM during heat treatment would allow improvement of the mechanical mechanical of these alloys.

In the present work, a Ti–26at.%Nb alloy was prepared by directly melting a mixture of elemental Ti and Nb powders under a laser beam. The alloy was annealed at temperatures ranging from 650 ◦C to 925 ◦C and the influence of the annealing temperature on the microstructure evolution and mechanical properties of the alloy was investigated.

#### **2. Experimental Methods**

#### *2.1. Material Manufacturing*

Commercial gas-atomized pure Ti powder (purity 99.99%, Changsha Tianjiu Metal Materials Co. Ltd., Changsha, China) with a particle size of 45–105 μm and pure Nb powder (purity 99.99%, Changsha Tianjiu Metal Materials Co. Ltd., Changsha, China) with a particle size of 48–75 μm were used as the raw materials. The two powders were mixed in the weight ratio Ti:Nb = 59.5:40.5 (namely 74:26 in atomic ratio) by ball milling for 1 h. Figure 1 shows the morphologies of the Ti and Nb powders and the mixtures. It is clear that the Ti powder has regular spheres with the presence of satellites (Figure 1a), whereas the Nb has irregular shapes (Figure 1b). After mixing, the Nb particles were uniformly distributed around the Ti particles (see Figure 1c). The melting point of Nb (2477 ◦C) is much higher than that of Ti (1668 ◦C). To facilitate better melting of the Nb and its faster diffusion in Ti, Nb powder with a smaller particle size than the Ti powder was used in the present work.

**Figure 1.** Scanning electron microscope (SEM) images of feedstock powders: (**a**) pure Ti; (**b**) pure Nb; and (**c**) Ti–26Nb powder mixture.

A square block sample of the obtained Ti–26Nb alloy with dimensions of 400 mm × 350 mm × 230 mm (see Figure 2) was fabricated by a LAM process on a YLS-4000-CL machine (IPG photonics corporation, Oxford, MA, USA), in which a fiber laser was produced by the IPG photonics corporation and the powder was fed in coaxial feeding mode with argon as the carrier gas. SD, LD, and BD represent the scanning direction, lateral direction, and build direction, respectively. The laser-deposition processing parameters were: laser nominal output power 750 W; laser beam diameter 2.5 mm; scanning speed 480 mm/min; and powder feed rate 2.2 g/min.

**Figure 2.** Bulk Ti–26Nb sample fabricated by laser additive manufacturing (LAM).

#### *2.2. Heat Treatment*

The samples for heat treatment were sectioned along the BD with dimensions of 25 mm in length (BD), 10 mm in width (SD), and 1.2 mm in thickness (LD). The samples were vacuum encapsulated in quartz tubes and annealed at temperatures ranging from 650 ◦C to 925 ◦C for a duration of 0.5 h followed by water quenching, as listed in Table 1.

**Table 1.** Details of heat treatment for LAM Ti–26Nb alloys.


Note: WQ is water quenching.

#### *2.3. Microstructure Characterization*

The microstructures of the alloy samples before and after heat treatment were characterized by optical microscopy (OM; BX51M, Olympus, Tokyo, Japan), scanning electron microscopy (SEM; MIRA3 LMU, Tescan, Brno, Czech), and X-ray diffraction (XRD; D/max 2500, Rigaku, Tokyo, Japan). Grain size and grain orientation were determined by electron backscatter diffraction (EBSD; HKL, Oxford, UK). Several regions in the samples were chosen for the EBSD analysis, and average grain size and grain orientations were determined by statistical analysis.

#### *2.4. Mechanical Property Testing*

Tensile tests of the Ti–26Nb alloy were carried out on an Instron 5569 universal testing machine (Instron, Boston, MA, USA). The samples for the tests were cut from the middle section of the block prepared by LAM with a gauge section of 1 mm × 2.5 mm × 8 mm, the geometry of which is schematically shown in Figure 3. All specimens are shown along the BD. Three samples were measured for each condition in order to reduce measurement error.

**Figure 3.** Schematic geometry of tensile specimen.

For comparison, another Ti–26Nb alloy was prepared using a conventional arc melting method with pure Ti and Nb as the raw materials. The ingot was hot-rolled by 90% in thickness, then tensile testing of the mechanical properties according to the above method was also carried out in the same conditions.

#### **3. Results and Discussion**

#### *3.1. Phase Composition*

Figure 4 shows the X-ray diffraction (XRD) patterns of the Ti–26Nb alloy fabricated by LAM and after annealing at different temperatures for 0.5 h. It can be seen that the as-deposited alloy shows a single β phase with a body-centered cubic (bcc) structure. After annealing at 650 ◦C for 0.5 h, diffraction peaks from an α phase could be seen in the XRD pattern of the alloy, implying that an α phase precipitated in the alloy. As the annealing temperature increased to 925 ◦C, the peaks from the α phase completely disappeared. By careful comparison, it can be seen that the diffraction peaks of the β phase on the XRD pattern of the alloy after annealing have slightly shifted to the left relative to the as-deposited sample (as shown in Figure 4). This implies that more Nb atoms have dissolved in the Ti lattice. This is because the radius of the solute atom Nb is slightly larger than that of the Ti atom, and thus a higher concentration of Nb leads to an increase in the lattice constant of the β phase.

**Figure 4.** X-ray diffraction (XRD) patterns of LAM-processed Ti–26Nb alloys after annealing treatment at different temperatures.

#### *3.2. Microstructure*

Figure 5 illustrates the microstructure of the as-deposited Ti–26Nb alloy. It is clear that multilayer deposits with uniform thickness and regular distribution have been formed in the as-deposited Ti–26Nb alloy. The prior β phase with columnar grains oriented more or less in the BD, which penetrate the multilayer cladding layer, can be observed (see Figure 5a). The special formation mechanism of columnar grains has been clearly explained in previous work [21]. Some unmelted Nb can be observed in the Ti–Nb alloy due to its relatively high melting temperature of 2477 ◦C, as shown in Figure 5a. The results indicate that the applied energy density of the LAM process is sufficient to completely melt the Ti powder, but some of the larger Nb particles are only partially melted. Similar behavior was also found in a previous report [22]. Moreover, some pores can be seen on the surface of the sample, which may have been caused by ball formation and gas inclusion generated during the melting and remelting process [23].

**Figure 5.** (**a**) and (**b**) Microstructures of as-deposited Ti–26Nb alloy; (**c**) and (**d**) elemental mapping images of as-deposited Ti–26Nb of (**b**).

A magnified image reveals that some ultra-fine elliptical-shaped grains (dendritic grains) with dimensions of ~10 μm in width and ~80 μm in length were formed in the representative regions (see Figure 5b). This may be attributed to the fact that the solidified layer acts as a substrate for the solidification of the melt, leading to the formation of elliptical-shaped grains perpendicular to the solidification front [24]. Moreover, the elliptical grains grew in a wavelike fashion and their boundaries are poor in Nb, as seen in Figure 5c,d.

Figure 6 shows the distributions of Ti and Nb elements in the molten pool of Ti-26Nb alloy. It can be seen that the Ti and Nb content at the melt pool boundaries (MPBs) is identical to that inside of the molten pool, implying that the solidification of the molten pool is completed in a short time, which is very different from conventional casting. As a result, the Nb particles do not have enough time to sink to the bottom of the molten pool and thus remain.

**Figure 6.** (**a**) Energy-dispersive X-ray spectroscopy (EDS) line scanning of melt pool in lateral plane of Ti–26Nb alloy (**b**) corresponding line scanning results of Ti and Nb elements.

To investigate the effects of the heat treatment on the microstructure of the alloy fabricated by LAM, the as-deposited alloy samples were annealed at temperatures ranging from 650 ◦C to 925 ◦C. Figure 7 illustrates the microstructures of the samples after annealing at the different temperatures.

For the sample annealed at 650 ◦C for 0.5 h, elliptical grains began to grow and their size distribution became more uniform (as shown in Figure 7a) as compared to as-deposited sample. Close observation reveals that the some fine acicular secondary α<sup>S</sup> phases precipitated in the areas poor in Nb atoms in the alloys annealed at 650 ◦C and 750 ◦C, as shown in the inset of Figure 7c,d). As the annealed temperature increased to 850 ◦C, the boundaries of the ellipses became unclear, implying the dissolution of the dendritic grains due to the diffusion of Nb atoms. Moreover, almost all the α<sup>S</sup> phases disappeared in the alloys as the annealing temperature increased over 850 ◦C, and the alloy that annealed at 925 ◦C exhibited an even microstructure with all β phases (see Figure 7e,f). This is consistented with the XRD analysis.

Figure 8 illustrates the Nb concentration in the elliptical-shaped grains and at the grain boundaries as obtained by energy-dispersive X-ray spectroscopy (EDS) analysis. It can be seen that, for the as-deposited alloy, the Nb content at the grain boundaries area is much lower than in the elliptical grains. After annealing at 650 ◦C, the Nb content at the grain boundaries increased to become close to that inside the grains, implying the diffusion of Nb from inside to the grain boundaries during the annealing treatment. As the annealing temperature further increased, the Nb content both inside and at the boundaries of the grains slightly increased, implying the dissolution of the unmelted Nb particles.

**Figure 7.** Microstructures of annealed samples at different temperatures for 0.5 h: (**a**) 650 ◦C; (**b**) 750 ◦C; (**c**) and (**d**) corresponding high-magnification morphologies of (**a**) and (**b**), respectively; (**e**) 850 ◦C; (**f**) 925 ◦C.

**Figure 8.** Comparison of Nb content in different regions of Ti–26Nb alloys after annealing at different temperatures.

#### *3.3. Grain Orientation and Grain Size*

As mentioned above, the microstructure of the as-deposited Ti–26Nb alloy exhibited columnar β grains with fine dendrites inside oriented more or less in the BD. The preferred direction of the dendrite growth in the Ti–Nb alloy can be ascribed to the largest thermal gradient parallel to BD during the LAM process, leading to the formation of this texture [25]. The texture was further verified by EBSD orientation maps, as seen in Figure 9a. It is clear that the coarse columnar β grains of the as-deposited alloy exhibits a relatively strong <100> fiber texture. The annealing treatment tended to make the grain orientation uniform, as shown in Figure 9b,c.

**Figure 9.** IPFZ electron backscatter diffraction (EBSD) map presented with respect to z-direction showing multiple grain orientations: (**a**) as-deposited Ti–26Nb alloy; (**b**) after annealing at 650 ◦C; (**c**) after annealing at 850 ◦C.

Further confirmation of the preferred orientation is illustrated by the corresponding EBSD pole figure in Figure 10. It can be seen that the as-deposited alloy exhibits the strongest <001> texture. After annealing treatment at 650 ◦C, the <001> texture of the alloy was weakened. However, with the annealing temperature further increasing to 850 ◦C, the texture became stronger.

**Figure 10.** EBSD pole figure and inverse pole figure of Ti-26Nb samples: (**a**) and (**d**) as-deposited Ti–26Nb alloy; (**b**) and (**e**) after annealing at 650 ◦C; (**c**) and (**f**) after annealing at 850 ◦C.

The effects of annealing temperature on the columnar grain size of the alloy were also investigated. Figure 11 shows the grain size distribution of the alloy before and after annealing at different temperatures. It can be seen that the average grain size of the as-deposited alloy is about 143.3 μm.

**Figure 11.** Grain size distributions of specimens: (**a**) as-deposited Ti–26Nb alloy; (**b**) after annealing at 650 ◦C; (**c**) after annealing at 750 ◦C; (**d**) after annealing at 850 ◦C; (**e**) after annealing at 925 ◦C.

While annealing treatment leads to clear growth in the grain size of the columnar β grains, with an increase in annealing temperature the size of the columnar β grains increases. The alloy after annealing at 925 ◦C exhibited the largest grain size, which reached 205.5 μm. The growth in grain size caused by the heat treatment may have decreased the strength of the alloy.

#### *3.4. Mechanical Properties*

The stress-strain curves of Ti–26Nb alloys after annealing at different temperatures (see Figure 12). The mechanical properties of the alloy before and after annealing at different temperatures were evaluated by Figure 12 and the results are shown in Figure 13 and Table 2. It can be seen that the as-deposited Ti–26Nb sample shows relatively high strength, with the ultimate tensile strength (UTS) and yield strength (YS) about 799 MPa and 768 MPa, respectively, which are much higher than those of the as-rolled Ti–26Nb alloy (see Table 2). The fine-grained structures present inside the β grains produced by the rapid solidification process may be responsible for the high strength [26].

**Figure 12.** Stress-strain curves of Ti–26Nb alloys after annealing at different temperatures.

**Figure 13.** Mechanical properties of Ti–26Nb alloys after annealing at different temperatures.


**Table 2.** Mechanical properties of Ti–Nb alloys fabricated by LAM and casting.

The heat treatment decreased the strength but increased the ductility of the alloy, depending on the annealing temperature. It can be seen that, after annealing at 650 ◦C for 0.5 h, the UTS and YS of the alloy reduced by about 176 and 223 MPa, respectively, in comparison with the as-deposited alloy due to the growth of the prior columnar β grains and the dissolution of the fine dendrites [26]. It should be noted that, as the annealing temperature further increased, the strength of the alloys gradually increased despite the growth of the grains. The UTS and YS of the alloy annealed at 925 ◦C were about 722 MPa and 685 MPa, respectively. The increase in strength of the alloy annealed at a relatively high temperature may be attributable to the solid solution strengthening the effect of the Nb solute atoms. As mentioned above, annealing treatment can lead to the dissolution of unmelted Nb atoms into the Ti lattice and, with an increase in annealing temperature, the content of Nb in the β increases. This may cause more severe lattice distortion, which can increase the resistance of the dislocation motion and consequently promote the strength of the alloy.

With regard to the ductility of the alloy, the as-deposited alloy exhibited relatively low ductility with elongation of 14.3%. This may be attributable to the presence of a large amount of residual stress in the as-built condition and defects such as pores, micro-segregation, and unmelted Nb particles in the as-deposited alloy. After annealing at 650 ◦C, the ductility of the alloy slightly increased due to the elimination of residual stress [27] and microstructural homogenization. However, the precipitation of secondary α phases at the grain boundaries may have deteriorated the ductility of the alloy. As the annealing temperature increased to 850 ◦C, most unmelted Nb particles dissolved into the β phase, which may have increased its stability, and thus no precipitation of secondary α phases occurred. This led to an increase in the ductility of the alloy. As a result, the alloy annealed at 850 ◦C exhibited relatively high ductility with elongation of 21.6%. However, with the annealing temperature further increasing to 925 ◦C, the grains of the β phase became coarse, which may have decreased the ductility of the alloy, as its elongation decreased to 18.9%.

#### **4. Conclusions**

In this work, a Ti–26Nb alloy was successfully prepared from a mixture of titanium and niobium powders by laser additive manufacturing. The as-deposited alloy was annealed at different temperatures ranging from 650 ◦C to 925 ◦C, then the microstructure evolution and tensile mechanical properties were evaluated. The main conclusions can be summarized as follows:

(1) The microstructure of the as-deposited Ti–26Nb alloy was characterized by prior columnar β grains with a relatively strong <001> fiber texture due to a large temperature gradient and remelting penetration in the build direction. Defects such as pores, unmelted Nb particles, and micro-segregation at the grain boundaries of the elliptical-shaped grains could be observed in the as-deposited Ti–26Nb alloy. Its ultimate tensile strength and yield strength were about 799 MPa and 768 MPa, respectively, much higher than those of hot-rolled Ti–26Nb (428 MPa and 415 MPa, respectively).

(2) The heat treatment had an important influence on the microstructures of the as-deposited alloy, depending on the annealing temperature. After annealing at 650 ◦C for 0.5 h, many fine acicular secondary α<sup>S</sup> phases precipitated at the boundaries of the elliptical-shaped grains with poor Nb content. With the annealing temperature increasing, the unmelted Nb particles dissolved into the Ti lattice, leading to an increase in Nb concentration in the β matrix, and the composition of the alloy tended to become uniform by the diffusion of Nb atoms. Furthermore, the annealing treatment led to growth in the size of the grains.

(3) Heat treatment decreased the strength but increased the ductility of the alloy. After annealing at 650 ◦C for 0.5 h, the prior columnar β grain growth and fine dendrite dissolution resulted in the ultimate tensile strength and yield strength decreasing to 623 MPa and 543 MPa, respectively. The increase in annealing temperature slightly increased the strength of the alloy. The solution strengthening due to the dissolution of the unmelted Nb particles together with the coarsening of the β grains may be responsible for the changing trend of the mechanical properties of the alloy with annealing temperature. The alloy annealed at 850 ◦C for 0.5 h exhibited a good balance of strength and ductility.

**Author Contributions:** J.W., J.L. and D.Z. designed experiments; J.W. and H.S. carried out experiments; J.L., J.W. and L.G. analyzed experimental results. J.L., D.Z. and C.W. assisted with Illumine sequencing. J.W. wrote the manuscript. J.L. and C.W. revised the paper.

**Funding:** This research was financially supported by the National Natural Science Foundation of China (11402220, 11872053), Scientific Research Fund of Hunan Provincial Educational Department (2016JC2005) and Hunan Provincial Natural Science Foundation of China (2018JJ4053).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Assay of Secondary Anisotropy in Additively Manufactured Alloys for Dental Applications**

#### **Elena Bassoli \* and Lucia Denti**

Department of Engineering "Enzo Ferrari", University of Modena and Reggio Emilia, via Pietro Vivarelli 10, 41125 Modena, Italy; lucia.denti@unimore.it

**\*** Correspondence: elena.bassoli@unimore.it; Tel.: +39-059-2056252

Received: 22 August 2018; Accepted: 19 September 2018; Published: 26 September 2018

**Abstract:** Even though additive manufacturing (AM) techniques have been available since the late 1980s, their application in medicine is still striving to gain full acceptance. For the production of dental implants, the use of AM allows to save time and costs, but also to ensure closer dimensional tolerances and higher repeatability, as compared to traditional manual processes. Among the several AM solutions, Laser Powder Bed Fusion (L-PBF) is the most appropriate for the production of metal prostheses. The target of this paper was to investigate the mechanical and microstructural characteristics of Co–Cr–Mo and Ti–6Al–4V alloys processed by L-PBF, with a specific focus on secondary anisotropy that is usually disregarded in the literature. Tensile specimens were built in the EOSINT-M270 machine, along different orientations perpendicular to the growth direction. Density, hardness, and tensile properties were measured and the results combined with microstructural and fractographic examination. For both alloys, the results provided evidence of high strength and hardness, combined with outstanding elongation and full densification. Extremely fine microstructures were observed, sufficient to account for the good mechanical response. Statistical analysis of the mechanical properties allowed to attest the substantial absence of secondary anisotropy. The result was corroborated by the observations of the microstructures and of the failure modes. Overall, the two alloys proved to be high-performing, in very close agreement with the values reported in the datasheets, independently of the build orientation.

**Keywords:** Powder Bed Fusion; Titanium alloys; Cobalt–Chrome alloys; anisotropy

#### **1. Introduction**

Additive Manufacturing (AM) techniques allow the production of objects with complex geometry. Fabrication can be started straightforwardly by using a three-dimensional Computer Aided Design (CAD) model, without tools. The basic idea is to think of every object as consisting of thin layers, usually in the range of 0.03–0.05 mm. The part is built up by progressive addition of material, which enables unprecedented ease of manufacturing of extremely complex shapes, since the three-dimensional manufacturing issues are simplified to two-dimensional problems [1,2]. It becomes, thus, possible to produce parts with cavities and undercuts that, by conventional subtractive methods, would have been unfeasible or would have caused great manufacturing hurdles and costs. AM technologies were introduced in industry in the late 1980s to realize models and prototypes, but nowadays the advances in materials and technology are sufficient to make the production of end products of major interest [3,4]. The great potential and good evolution of techniques led to introduce AM in medicine, where the need for parts that are customized for each patient, with a high degree of personalization, allows full exploitation of the inherent benefits of additive processes [5]. In particular, the Laser-based Powder Bed Fusion (L-PBF) process can be successfully used for the production of prostheses [6], including for example long-span and cantilever metal-ceramic-fixed partial dentures for

maxillary and mandibular prosthodontics [7,8]. Apart from restorations, also surgical guides aimed at operation planning can take advantage of the quick production by AM [6,9].

This paper focuses on the characterization of alloys developed on purpose for maxillo-facial surgery, in particular for oral implants. Implantology is a field under continuous innovation, where research efforts are concurrently dedicated to finding new materials, new components, new fabrication processes, with the aim to improve the duration, the aesthetics and the functionality of prostheses and thus ensuring a better quality of life for patients. An easy example is that of gold crowns that have been superseded by metal-ceramic or ceramic crowns [10–12]. An increasing variety of metallic biomaterials is being developed, ranging from commercially pure Titanium [13] and Titanium alloys, through stainless steels, to Cobalt–Chrome alloys. Promising results have recently been attained for innovative β-type Ti alloys with increased wear resistance and lower elastic modulus, so as to better match that of the human bone and prevent the stress shielding effect [14]. Some of these novel alloys exhibit composite microstructures where β–Ti dendrites are surrounded by intermetallic phases so that efficient reinforcing phenomena are established [15]. The promising outlook of these Titanium-based matrix composites for AM has been very recently reviewed [16].

L-PBF process can be used in the construction of metal-ceramic fixed partial dentures (FPDs). Traditionally, the manufacture of the metallic part of FPDs involves a large series of manual operations performed by the dental technician, and the result is often strongly determined by his ability. For metal-prostheses fabrication, the adoption of L-PBF grants a much higher repeatability and predictability with respect to the manual process [17]. Compared to other powder-based methods that require molds, PBF offers outstanding personalization capabilities, in extremely short times and with low costs. On the other hand, the L-PBF process is quite complex and many factors are involved in order to achieve good part quality [18]. Despite the extremely diffused studies on this process, many efforts are still needed to better understand the relation between microstructure, processing, and properties for parts built by L-PBF [19–21].

In L-PBF, at each step, a thin layer of metallic powder is evenly distributed onto the previous layer and a laser selectively scans the regions corresponding to the cross-section of the part. As a consequence, the powder melts and then consolidates into a solid slice. Inherent in the process are two types of possible anisotropy: a primary one, due to the superimposition of layers in the direction that is usually called Z; and a secondary one that may manifest as direction-dependence of properties even within the XY plane, that is to say parallel to the layers [22]. The latter is usually ignored by machine- and material suppliers, and has been disregarded by scientific literature until now. Secondary anisotropy may be caused for example by the action of the recoater blade that spreads the powder in the bed, or by the inert gas flux that blows the melting slags away from the build area [23]. Each of the two phenomena usually acts along either the X or the Y direction, depending on the specific machine architecture. The investigation of secondary anisotropy is markedly important if the intended application is the production of FPDs, because complete prostheses are not straight structures but develop along the maxillary/mandibular arch, hence they involve material properties in several directions of the XY plane. A robust design of the restoration requires a reliable knowledge of any direction-dependent feature.

This research tackles the mechanical properties and the microstructure of two L-PBF fabricated dental alloys, namely Co–Cr–Mo and Ti–6Al–4V, by proposing a statistically-based enquiry of secondary anisotropy.

#### **2. Materials and Methods**

Tensile specimens were produced by L-PBF using the two alloys Ti–6Al–4V (EOS GmbH, Krailling, Germany) and Co–Cr–Mo (EOS Cobalt Chrome MP1, EOS GmbH, Krailling, Germany).

The specimens were fabricated on the L-PBF machine EOSINT-M270, by using the following process parameters:

• for Ti–6Al–4V: laser power 340 W, laser spot diameter 0.1 mm, layer thickness 30 μm, scan speed 1250 mm/s, hatch distance 0.12 mm, protective atmosphere (max 0.1% oxygen);

• for Co–Cr–Mo: laser power 200 W, laser spot diameter 0.2 mm, layer thickness 20 μm, scan speed 7000 mm/s, hatch distance 0.3 mm, protective atmosphere (max 1.5% oxygen).

For both alloys, tensile specimens were built in three different orientations relative to the machine distinctive directions, all of the three parallel to the layers and perpendicular to the growth direction. The three groups, each of 6 specimens, are specified as follows:


The size and geometry for the tensile test specimens conformed to the prescription specified in standard ASTM E8M [24]. Details are reported in Figure 1.

Of the powders, the nominal physical/mechanical properties and chemical composition are listed in Tables 1 and 2, respectively. The powders were characterized by means of laser granulometry (Malvern Mastersizer 3000, Malvern Panalytical Ltd., Malvern, UK) to assess their size distribution, according to ISO 13320 standard [25].

**Figure 1.** Tensile specimen. Dimensions are expressed in millimeters.

**Table 1.** Nominal physical and mechanical properties of the two alloys.


\* in horizontal direction, as built condition.

**Table 2.** Nominal chemical composition of the two alloys.


The specimens were tested in the as built condition, without any heat treatment, so as to avoid any smoothing of the secondary anisotropy produced by the L-PBF process.

Before the tensile tests, the Archimedes principle was used to measure the density of all the samples (6 for each group), with an analytical electronic balance having a resolution of 0.1 mg (Pioneer textsuperscript® Plus PA124C, OHAUS GmbH, Greifensee, Switzerland). The residual porosity was then calculated by using the nominal density of each alloy.

Tensile tests were performed on a SCHENK HYDROPULS PSB testing machine (SCHENCK RoTec GmbH, Darmstadt, Germany) with a capacity of 250 kN, using a crosshead speed of 5 mm/min. Five samples were tested for each alloy and orientation, and one extra specimen of each group was used to measure hardness and to obtain the metallographic sections. The choice of the hardness scale was made according to ISO standard 4498 [28]. Rockwell C was selected and performed following the specifications of standard ISO 6508 [29], by repeating five measurements on each sample. Numerical results for hardness (HRC), tensile strength (UTS) and total extension at fracture (εb) were processed through statistical tools (Statistica 8, Statsoft, Hamburg, Germany): the *t*-test with a level of significance of 0.05 was performed to investigate the presence of significant differences between the groups of specimens produced along different orientations.

After tensile tests, rupture surfaces were observed by using a scanning electron microscope, SEM (ESEM, Quanta FEI, Thermo Fisher Scientific, Eindhoven, The Netherlands), in order to investigate the failure mechanisms and the joining phenomena between the particles.

Metallographic sections of the samples were obtained and observed by an optical microscope (OM) (Eclipse LV150N, Nikon, Tokyo, Japan), to get a cross-check of residual porosity and compare the results with those obtained by the Archimedes method. A comparative assessment of the two methods is raising the interests of the scientific community [30], growingly as the two techniques are more and more diffused in industry for the control of AM parts. Preparation of the metallographic sections consisted of micro-cutting, embedding in epoxy resin and polishing till a fine grinding. The final step was carried out with a plan cloth and 1 μm diamond suspension. Several micrographs were acquired through a CCD camera, made binary and analyzed through a software tool for image analysis to determine:


After OM observation, polished sections of Ti–6Al–4V underwent chemical etching with the Dix-Keller reactant (HF 2% vol, HCl 1.5% vol, HNO3 2.5% vol; water bal.); while metallographic sections of the Co–Cr–Mo alloy were subjected to electrochemical etching (HCl 0.1 M, 2 V, 2 min). Microstructures were observed on the etched samples by means of OM and SEM.

#### **3. Results and Discussion**

#### *3.1. Powder Particle Size Distribution*

The results of laser granulometry of the two powders are shown in Figure 2. Co–Cr–Mo powder displays a wider distribution, with an average value of the order of 80 μm and a relatively large number of particles in the range 10–30 μm causing a negative skew in the curve. The size distribution for Ti–6Al–4V powder is instead symmetrical, with an average particle dimension of 30 μm.

**Figure 2.** Particle size distribution of the Co–Cr–Mo (**A**) and Ti–6Al–4V (**B**) powders.

#### *3.2. Density and Residual Porosity*

Table 3 lists the results for density and for residual porosity measured by the Archimedes principle, as well as by OM observations of metallographic sections. Figures 3 and 4 show, respectively for Ti–6Al–4V and Co–Cr–Mo, examples of the OM images on which residual porosity was calculated by image analysis. The nominal densities of the two alloys are available in Table 1 for comparison. While the data obtained by the Archimedes method are normally distributed, the data by microscopic analysis exhibit the asymmetrical distributions shown in Figure 5. Hence, mean values and standard deviations are listed in Table 4 for the Archimedes figures, whereas median and mean values are reported for the analysis of OM images. Density figures are, for both alloys, very close to the nominal values, with extremely narrow deviations and no evident direction dependence, as the differences between the values calculated for the X, Y, and XY groups are contained within the standard deviations. Residual porosity is in all cases well below 1%, with no distinction for the various orientations. If porosity is calculated by comparing the Archimedes density with the nominal one, the values are slightly higher than those obtained by metallographic observations, with the only exception of Ti–6Al–4V X specimens. Based on these results, the Archimedes method can be reckoned conservative if applied to density control of L-PBF fabricated parts. This remark is in very good agreement with the results attained by Spierings et al. [30], who found the Archimedes measurement highly accurate and repeatable for the control of metal parts produced by PBF. The same study also concludes that, in contrast, microscopic analysis of cross sections can give inconsistent values of density, with variations of up to 4% in the direction of an underestimate of the residual porosity. The two methods are found comparable by Spierings et al. only for low porosities.


**Table 3.** Density and porosity determined by the Archimedes principle and by OM observations. For Archimedes measurements, standard deviations are given in brackets next to the mean values computed over 6 measurements. For the results obtained by OM, median and mean values are provided.

**Table 4.** Results of the hardness and tensile tests. Standard deviations are given in brackets next to the mean values.


**Figure 3.** OM images of metallographic sections of Ti–6Al–4V specimens of the X (**A**), Y (**B**), and XY (**C**) groups.

**Figure 4.** OM images of metallographic sections of Co–Cr–Mo specimens of the X (**A**), Y (**B**), and XY (**C**) groups.

**Figure 5.** Distribution of the area of pores detected by the analysis of OM images: (**A**) Ti–6Al–4V, (**B**) Co–Cr–Mo.

#### *3.3. Hardness and Tensile Tests*

The results of hardness and tensile tests are listed in Table 4. Representative tensile stress–strain curves are shown in Figure 6. Necking is observed for Ti–6Al–4V, whereas the Co–Cr–Mo graphs are bilinear. As a term of comparison, mechanical properties reported in literature for the same alloys are given in Table 5. Normality of data distribution was verified by using the Shapiro-Wilk test, for all the mechanical characteristics recorded in Table 4. Then, the *t*-test was performed by grouping the mechanical properties according to the variable "orientation". Table 6 registers the results, expressed in terms of probability values (*p*-values). When lower than 0.05, the *p*-values can be taken as a decision to reject the null hypothesis of absence of significant differences between the groups, that is to say of absence of anisotropy. In other terms, when the *p*-value is lower than 0.05 the mechanical response in different orientations can be regarded as non-equivalent.

**Table 5.** Tensile properties reported in literature for the Ti–6Al–4V and the Co–Cr–Mo alloys.



**Table 6.** Values resulting from the *t*-test for the variables HRC, UTS, ε<sup>b</sup> among the groups built in different orientations. Records below the level of significance of 0.05 are bold.

**Figure 6.** Example stress–strain curves of the Ti–6Al–4V (**A**) and Co–Cr–Mo (**B**) specimens built in the three orientations.

HRC hardness is nearly 39 for Titanium alloy samples and 47 for Co–Cr–Mo specimens, with no statistically-significant differences between samples produced in different orientations. Mean UTS of Titanium specimens resulted in the range 1080–1110 MPa, and extension is about 12.5%. As for the Co–Cr–Mo alloy, mean UTS is between 1280 and 1300 MPa and values of 13–14%were obtained for total extension at fracture. All groups exhibit good test repeatability, with very low standard deviations, except for Ti–6Al–4V XY samples that show a relatively high scattering of the measured UTS. A slight anisotropy can be noticed for Ti–6Al–4V, consisting of a decreasing trend of strength from the X- towards the Y-orientation. Nevertheless, the variation is proportionately low (2.8%) and is pointed out as significant by the *t*-test only when two extreme groups, X and Y, are compared. A similar consideration is valid for Co–Cr–Mo samples, but in this case the opposite trend is observed: strength increases as the build orientation varies from the X- to the Y-direction. Even if the *t*-test is positive in two of the three cases, yet the overall deviation is as little as 1.6%. From a practical standpoint, for any industrial application the mechanical properties in the three directions would be considered undifferentiated, as a deviation of few per cents is by far absorbed by the factor of safety. As to ductility, for all the specimens the values are remarkably high if compared either to the nominal characteristics or to the typical properties of L-PBF fabricated parts. Furthermore, no direction dependence is evidenced for total extension at fracture.

#### *3.4. Fractography*

For Ti–6Al–4V specimens, failure occurs by a variety of mechanisms that can be observed comprehensively on macro-views of the rupture surfaces, as in the example in Figure 7A. The rupture shown is a typical failure mode of ductile materials, usually designated as cup and cone rupture. This form of ductile failure begins after necking and develops through sequential steps. At the outset, small micro-voids appear in the innermost zone of the specimen. Then, as plastic deformation proceeds, the micro-voids expand and merge into a crack. Figure 7B,C allow to appreciate, at high magnifications, the failure morphologies of the areas that are marked in Figure 7A as "B" and "C", respectively. As to the first, laterally-fine but raised dimples can be observed, with a large amount of plastic deformation. This morphology can also be designated as cellular fracture and is often detected on the rupture surfaces of additively manufactured multi-phase materials [36]. On the last-breaking areas, as in Figure 7C, dimples are much flatter, indicating a low energy dissipation. Figure 7D shows a zone of transition between the two described failure modes. Lack of fusion defects are exceptionally spotted on the rupture surfaces (Figure 7E) [24]. Failure modes and rupture morphologies are identical for the specimens produced in the three orientations, as can be reckoned by comparing Figure 7 (X specimen) with Figure 8 (Y and XY specimens).

Co–Cr–Mo specimens exhibit a homogeneous morphology across the rupture surfaces. A representative case for each of the three build orientations is shown in Figure 9, where failure seems to occur mainly by transgranular cleavage, even if the values of extension at fracture would suggest a more ductile mode. In these cases, the term "quasi-cleavage" is usually adopted, to identify a rupture that combines cleavage-like features with evidence of plastic deformation. Also, in this case, as for Ti–6Al–4V, the rupture surfaces of the three groups of samples are totally equal to each other.

**Figure 7.** SEM observations of the rupture surface of a Ti–6Al–4V X specimen: (**A**) overall view; (**B**) dimpled rupture morphology; (**C**) quasi-flat rupture morphology; (**D**) mixed morphology; (**E**) lack of fusion defect.

**Figure 8.** SEM observations of the rupture surfaces of a Ti–6Al–4V Y (**A**,**B**) and XY (**C**,**D**): (**A**,**C**) quasi-flat rupture morphology; (**B**,**D**) dimpled rupture morphology.

For both alloys, if the different building orientations are compared, different rupture morphologies can be perceived in the macroscale, but the micro-mechanisms are in effect equal. This result is consistent with the substantial equivalence of the mechanical properties that has been discussed in Section 3.3. Overall, fractography suggests an extremely fine grain structure, which will be verified in the next section by means of observations of etched sections.

**Figure 9.** SEM observations of the rupture surface of Co–Cr–Mo specimens: (**A**,**B**) X; (**C**,**D**) Y; (**E**,**F**) XY.

*3.5. Microstructure*

The microstructure observed for Ti–6Al–4V samples in the different orientations is visible in Figure 10. The sections in Figure 9B,C and Figure 10A are perpendicular to the axes of a Y, X, and XY specimen, respectively. Figure 10D allows to appreciate the microstructure in the XY plane. The Ti–6Al–4V specimens exhibit a uniform acicular α martensite microstructure [13], which forms under elevated cooling rates. Rapid cooling is frequently reported in literature for

AM processes [19,37,38]. As an example, Criales et al. in a study on the L-PBF process of Inconel 625 measured cooling rates in the order of 150 ◦C/ms [38]. For Ti–6Al–4V, a rapid quenching is known in literature to cause a martensitic transformation, leading to a very fine needle-like microstructure [39]. No evidence of direction-dependent features is noticed.

Images of the etched sections of Co–Cr–Mo specimens, referred to all the orientations already shown for Ti–6Al–4V samples, are provided in Figure 11A,D. For this alloy, etching reveals the boundaries of the melt pools. In all directions, within each melt pool an ultra-fine columnar microstructure can be appreciated at high magnification (Figure 11E,F), with varying grain orientation. The columnar grains have sub-micron diameter. No evidence of a martensitic structure is visible in Figure 11, however these observations are not conclusive on the point, in the absence of specific investigations, as for example X-ray diffractions.

**Figure 10.** SEM images of etched sections of Ti–6Al–4V specimens. Alignment of the (**A**–**D**) images is schematized in the cube.

**Figure 11.** SEM images of etched sections of Co–Cr–Mo specimens. Alignment of the (**A**–**D**) images is schematized in the cube. (**E**,**F**) Detail of the columnar grains.

#### **4. Conclusions**

In view of the increasing interest in the use of additively manufactured parts for dental prostheses, the mechanical behavior and the microstructure of Ti–6Al–4V and Co–Cr–Mo parts, built by L-PBF, were investigated, with a specific focus on the evaluation of secondary anisotropy.

For both alloys, the measured hardness and strength were in good agreement with those reported in the datasheets; ductility was remarkably high and nearly full densification was measured. The observed microstructures, typical of the extreme cooling rates experienced by the materials during L-PBF processes, allow to account for the outstanding mechanical properties that were appraised in this study. Statistical analysis of the mechanical properties allowed to attest the substantial absence of secondary anisotropy and the result was confirmed by the observation of identical failure modes of the specimens produced in the different orientations.

On the whole, the results enable the conclusion that the two alloys considered here may achieve exceptionally high properties if manufactured by L-PBF, and that secondary anisotropy is negligible if not totally absent.

**Author Contributions:** Conceptualization, Methodology, Writing: Review and Editing E.B.; Validation, Investigation, Writing: Original Draft Preparation, L.D.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*
