Low Cycle Fatigue Performance of Additively Processed and Heat-Treated Ti-6Al-7Nb Alloy for Biomedical Applications

: In biomedical engineering, laser powder bed fusion is an advanced manufacturing technology, which enables, for example, the production of patient-customized implants with complex geome-tries. Ti-6Al-7Nb shows promising improvements, especially regarding biocompatibility, compared with other titanium alloys. The biocompatible features are investigated employing cytocompatibility and antibacterial examinations on Al 2 O 3 -blasted and untreated surfaces. The mechanical properties of additively manufactured Ti-6Al-7Nb are evaluated in as-built and heat-treated conditions. Recrystallization annealing (925 ◦ C for 4 h), β annealing (1050 ◦ C for 2 h), as well as stress relieving (600 ◦ C for 4 h) are applied. For microstructural investigation, scanning and transmission electron microscopy are performed. The different microstructures and the mechanical properties are compared. Mechanical behavior is determined based on quasi-static tensile tests and strain-controlled low cycle fatigue tests with total strain amplitudes ε A of 0.35%, 0.5%, and 0.8%. The as-built and stress-relieved conditions meet the mechanical demands for the tensile properties of the international standard ISO 5832-11. Based on the Cofﬁn–Manson–Basquin relation, fatigue strength and ductility coefﬁcients, as well as exponents, are determined to examine fatigue life for the different conditions. The stress-relieved condition exhibits, overall, the best properties regarding monotonic tensile and cyclic fatigue behavior.


Introduction
Biomedical materials often are used for replacing lost or diseased biological structures [1][2][3]. Implants require adequate properties, including mechanical properties, such as high wear resistance, corrosion resistance, excellent biocompatibility, osseointegration, and non-cytotoxicity, to avoid revision surgeries [4].
Titanium and its alloys are generally used in biomedical applications due to their excellent biocompatibility, high corrosion resistance, and superb mechanical properties, such as low elastic modulus and high strength. Titanium alloys are the most widely used metallic materials for load-bearing biomedical applications [5][6][7][8][9]. Ti-6Al-7Nb is an (α + β) titanium alloy with high specific strength and corrosion resistance, accompanied obtain dense material with LPBF (relative density ϕ > 99.9%), the following parameters were used: laser power P = 227 W, laser scanning speed v = 1.675 mm s −1 , hatch distance h = 0.077 mm. A constant scanning strategy of 5 mm stripes by layer-wise rotation of the scanning vectors of 67 • alongside a defined layer thickness of 50 µm was applied. As contour parameters, a laser power P c = 123 W and a scanning speed v c = 0.512 mm s −1 were applied. The Ti-6Al-7Nb powder is supplied by ECKART TLS GmbH (Bitterfeld, Germany). The Ti-6Al-7Nb powder particles were examined concerning particle size distribution (PSD) with a Mastersizer 2000 (Malvern Panalytical GmbH, Kassel, Germany) using laser diffraction. The powder batch had a nominal PSD comprising 26.9 µm (D 10 ) and 52.1 µm (D 90 ) with a log-transformed normal distribution centered at 37.6 µm (D 50 ), see Figure 1a. Different heat treatments were performed under a vacuum atmosphere. Recrystallization annealing (HT1 = 925 • C for 4 h) was conducted to obtain equiaxed α with β at grain boundary triple points, β annealing (HT2 = 1050 • C for 2 h) to receive Widmanstätten α + β colony structures, and stress relieving (HT3 = 600 • C for 4 h) to decrease the undesirable residual stresses due to the LPBF process, as well as a decomposition of the martensitic α to α-phase [59,60]. The heat treatment process routes are presented in Figure 1c. The powder morphology was investigated employing the scanning electron microscope (SEM) Zeiss Ultra Plus (Carl Zeiss AG, Oberkochen, Germany). The powder consists of mainly spherical particles with a few agglomerations on bigger particles, see Figure 1b. Microstructural investigations were accomplished with the SEM equipped with an electron backscatter diffraction (EBSD) unit to detect phases and corresponding grain orientations. EBSD data was post-processed using the MATLAB-based (Version R2019a 9.6, The MathWorks, Inc., Natick, MA, USA) toolbox MTEX (Version 5.6.0) [61]. MTEX is a free available toolbox for analyzing and modelling crystallographic textures. Microscopic observations and classification of the fractured surfaces for quasi-static and fatigue tests were also carried out using SEM. Microstructural study of samples on nanoscale was undertaken by transmission electron microscopy (TEM). For TEM, thin-slice samples (≈ 400 µm) were cut by Struers Sectom-5 (Struers GmbH, Willich, Germany) and further polished to a thickness of ≈ 100 µm. At last, 3 mm diameter circular disc samples were punched from the thin samples. Twin jet electropolishing of the thin foils was performed with the Struers Tenupol-5 (Struers GmbH, Willich, Germany) using an electrolyte containing perchloric acid (60 mL), butanol (340 mL), and methanol (600 mL). Electropolishing was executed at a voltage of 21 V, a current of 35 mA, and a temperature of −22 • C. TEM investigations were executed using a cold field emission gun equipped with JEOL JEM-ARM 200F (JOEL Ltd., Tokyo, Japan). TEM, high-resolution TEM (HRTEM), and high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images were taken. Energy dispersive spectroscopy (EDS) was performed during HAADF-STEM imaging. EDS maps were collected with a 30 nm step size with 10 s dwell time per step. The crystalline phase composition of Ti-6Al-7Nb was characterized through X-ray diffraction (XRD; D8 Advance, Bruker, Madison, WI, USA) using a Cu-Kα radiation source (λ = 0.154187 nm) operating at a current of 40 mA and a voltage of 35 kV. The as-built and heat-treated Ti-6Al-7Nb parts were measured in θ-θ geometry within a scanning range from 2θ = 30 • to 2θ = 90 • applying a step width of 0.035 • and an exposure time of 1.5 s per step. All tests for mechanical characterization were performed at an ambient temperature. Tensile specimens were built according to Figure 1d. The loading direction was parallel to the building direction (BD) of the samples. All tensile specimens were blasted with high-grade Al2O3. The blasting mat corresponded to a particle size of 70 µm-250 µm, used at 4 bar air pressure with a S 25 DUO (MHG Strahlanlagen GmbH, Düsseldorf, Germany). The tensile and LCF were performed utilizing a servo-hydraulic test-rig MTS 858 table-top system ( Systems Corporation, Eden Prairie, MN, USA) equipped with a 20 kN load cell an extensometer 632.29F-30 (MTS Systems Corporation, Eden Prairie, MN, USA). The te test procedure corresponded to a displacement-controlled execution with a crossh speed of1.5 mm min −1 , according to DIN EN ISO 6892-1. The LCF tests were st controlled, at total strain amplitudes εA of 0.35%, 0.5%, and 0.8%, with a R-ratio o (compression-tensile fatigue), and a strain rate of 6 × 10 −3 s −1 . At least three specimens condition were tested with both monotonic and cyclic tests for each strain amplitude. monotonic material properties, such as Young's moduli E, tensile yield strengths ultimate tensile strengths Rm, and plastic elongations A, were obtained from the s tensile test. The determination of the fatigue material constants requires the perform (c) schematic overview of the different heat treatments conducted-recrystallization annealing (HT1), β annealing (HT2), and stress relieving (HT3), as well as the β transus temperature for the alloy; (d) geometry and building direction (BD) of tensile and fatigue specimens.
All tensile specimens were blasted with high-grade Al 2 O 3 . The blasting material corresponded to a particle size of 70 µm-250 µm, used at 4 bar air pressure with a SMG 25 DUO (MHG Strahlanlagen GmbH, Düsseldorf, Germany). The tensile and LCF tests were performed utilizing a servo-hydraulic test-rig MTS 858 table-top system (MTS Systems Corporation, Eden Prairie, MN, USA) equipped with a 20 kN load cell and an extensometer 632.29F-30 (MTS Systems Corporation, Eden Prairie, MN, USA). The tensile test procedure corresponded to a displacement-controlled execution with a crosshead speed of1.5 mm min −1 , according to DIN EN ISO 6892-1. The LCF tests were strain-controlled, at total strain amplitudes ε A of 0.35%, 0.5%, and 0.8%, with a R-ratio of −1 (compression-tensile fatigue), and a strain rate of 6 × 10 −3 s −1 . At least three specimens per condition were tested with both monotonic and cyclic tests for each strain amplitude. The monotonic material properties, such as Young's moduli E, tensile yield strengths R p0,2 , ultimate tensile strengths R m , and plastic elongations A, were obtained from the static tensile test. The determination of the fatigue material constants requires the performance of several fatigue tests under cyclic loading and a R-ratio of −1. Cyclically loaded materials often have unstable and changing stress amplitude during the test, due to cyclic hardening or softening. Therefore, the stress amplitude for the stabilized state must be used, which occurs at the half number of cycles to fracture. The plastic strain amplitude can either be calculated with the stress amplitude, the Young s modulus and the total strain amplitude (see Equation (4)), or by measuring the thickness of the recorded stable hysteresis loop recorded during the fatigue tests [62][63][64][65]. For evaluation of the total stress amplitude, σ a , from the S-N curves, the Basquin relation is suitable, as follows: In Equation (1), σ f delineates the fatigue strength coefficient and b is the fatigue strength exponent. This equation fits the high-stress and low-stress fatigues [66]. For a better description of the high-stress fatigue region, the dependence between the number of reversals to failure 2N f and the plastic strain amplitude ∆ε p /2, also called the Coffin-Manson relation, is used to describe the total fatigue life, as follows: with the fatigue ductility coefficient, ε f , and the fatigue ductility exponent, c [66][67][68]. The total strain amplitude, ∆ε A /2, can be divided into two components, the plastic strain amplitude, ∆ε p /2, and the elastic strain amplitude, ∆ε e /2. The total strain amplitude, ∆ε/2, of fatigue life curves is described as the sum of elastic and plastic strain amplitude, as follows, by Suresh [69]: ∆ε e /2 is described by Hooke s law as the quotient of σ a to the Young s modulus E as follows: Together with the modified Basquin equation and the Coffin-Manson relation, combining Equations (1)-(4), one obtains the following: The first and second terms on the right-hand side of Equation (5) are the elastic ε e and plastic ε p components, respectively, of the total strain amplitude ε A . Equation (5) can be used as the basis for the strain life approach to fatigue design. For determination of the fatigue life, a schematic illustration of the as-built condition is presented in Figure 2. The intersection of the curves for ε e and ε p describes the transition point, where the plastic and elastic strains are identical. From this point, the LCF life is governed more by elastic than plastic strain. For the as-built condition this point is around 110 cycles, see Figure 2. According to the Coffin-Manson-Basquin approximation, one is able to predict the fatigue life depending on the applied strain amplitude for the different conditions [66][67][68]70].
intersection of the curves for εe and εp describes the transition point, where the plastic and elastic strains are identical. From this point, the LCF life is governed more by elastic than plastic strain. For the as-built condition this point is around 110 cycles, see Figure 2. According to the Coffin-Manson-Basquin approximation, one is able to predict the fatigue life depending on the applied strain amplitude for the different conditions [66][67][68]70].

Biocompatibility Investigations
Two different surface conditions were analyzed to determine the optimum surface roughness for biocompatibility. On the one hand, the as-received surface was investigated without any post-treatment (NT). On the other hand, the surface was post-treated utilizing blasting with high-grade Al 2 O 3 (ab). Blasting lead to rougher surfaces and to an increased surface area, resulting in faster osseointegration and higher survival rates for moderately rough implant surfaces: this was demonstrated by clinical studies [71][72][73][74][75]. The surface roughness was measured with an optical 3D macroscope VR-3100 (KEYENCE GmbH, Neu-Isenburg, Germany). The arithmetical mean roughness, Ra, and average roughness, Rz, were determined on the surfaces of the untreated and blasted specimens. A total of 5 specimens per condition were measured with 5 line measurements (length ≈ 6.8 mm) per side.

Cytocompatibility Testing
Biocompatibility studies were performed with different cell types. Therefore, murine fibroblasts (L-929; CLS Cell Lines Service, Eppelheim, Germany), human osteosarcoma cells (HOS 87070202; European Collection of Authenticated Cell Cultures, Merck, Germany), and human umbilical vein endothelial cells (HUVEC; Promocell, Heidelberg, Germany) were used. L-929 cells were grown and passaged in RPMI-1640 medium (Biochrom GmbH, Berlin, Germany), while HOS cells were grown and passaged in Eagle's MEM (EMEM)/Hanks' (Carl Roth GmbH + Co. KG, Karlsruhe, Germany). Both media were supplemented with fetal calf serum (FCS) superior (10%, Biochrom GmbH, Berlin, Germany) and penicillin/streptomycin (1%, Biochrom GmbH, Berlin, Germany). The EMEM/Hanks medium was also supplemented with non-essential amino acids (1%, Biochrom GmbH, Berlin, Germany) and L-glutamine (2 mmol, Biochrom GmbH, Berlin, Germany). HUVEC cells were grown and passaged in an endothelial cell grown medium kit (Promocell, Heidelberg, Germany) and were supplemented with penicillin/streptomycin (1%). All cell lines were grown and passaged in cell culture flasks or multi-well plates (Greiner Bio-One, Frickenhausen, Germany). For passaging, a trypsin (0.05%)/ethylene-diamine-tetraacetic (0.02%) acid solution (Biochrom GmbH, Berlin, Germany) was used. The cells were plated with a density of 50,000 cells per well in 24-well culture plates. After hot air sterilization of the test samples, the cells were seeded on the top of the samples (≈5 × 5 mm 2 ), which were embedded in agarose (2%, Agarose NEEO, Carl Roth GmbH + Co. KG, Karlsruhe, Germany) in respective culture mediums. After 72 h incubation (in a humidified atmosphere at 37 • C and 5% CO 2 ), cell proliferation was determined. Therefore, a crystal violet staining (CV) assay was performed [76]. After 72 h, the cells were fixed with glutaraldehyde (2%, Sigma-Aldrich, Merck KGaA, Darmstadt, Germany) in phosphate-buffered saline (PBS) for 20 min. Subsequently, the supernatant was removed, and the cells were stained with CV (0.1%, Carl Roth GmbH + Co. KG, Karlsruhe, Germany) in deionized water. After 30 min, the pigment was removed, and the samples were removed from the agarose. After washing with deionized water, the test samples were air-dried. For 1 h, Triton X-100 (2%, Sigma-Aldrich, Merck KGaA, Darmstadt, Germany) in deionized water was added, so that CV was dissolved from the cells. Finally, the supernatant was transferred in a 96-well microtiter plate and the absorbance was determined at 570 nm using a 96-well microplate reader (MRX microplate reader, Dynatech Laboratories, Denkendorf, Germany). The experiments were carried out six times. Furthermore, the supernatant of L-929 and HOS was analyzed of the cytokine Il-6 expression (R&D Systems DuoSet, R&D Systems Inc., Minneapolis, MN, USA).

Antibacterial Examinations
To analyze the bacterial behavior on test samples, two different samples were used-Escherichia coli (ATCC 25922) and Staphylococcus aureus (ATCC 25923). The bacteria were cultivated on Columbia agar plates with sheep blood (7%, Fisher Scientific GmbH, Schwerte, Germany) for 24 h. The test samples were also embedded in agarose (2%, Agarose LM, Gerbu Biotechnik GmbH, Gailberg, Germany). Bacteria (10 −6 CFU ml −1 ) were seeded on the test samples. After 48 h of incubation at 37 • C, cell proliferation was analyzed, similarly to the cell proliferation, with a CV assay. The experiment was performed in four biological replicates.

Microstructure and Nanostructure of as-Built and Heat-Treated Ti-6Al-7Nb
The microstructure of (α + β) titanium strongly depends on the cooling rates and the quenching parameters, from the β-phase field at higher temperatures and the following heat treatment, respectively [77,78]. Figure 3 shows EBSD maps of the Ti-6Al-7Nb alloy after additive manufacturing and subsequent heat treatments. The EBSD map of the additively processed Ti-6Al-7Nb shows a very fine, acicular microstructure, see Figure 3a,c. The fine-lamellar α grain structures are strongly oriented inside the prior β grains. The size of these fine needles decreased with an increased cooling rate during solidification [37,79,80]. Figure 3b shows the reconstructed parent β grains using MTEX software. During solidification, the bcc β-phase preferentially grows in the <100> direction; therefore, the elongated, columnar primary β grains evolve parallel to the BD [81][82][83]. The resulting anisotropy and primary β grains affect the mechanical properties of the specimens [84]. Due to the fast cooling and passing the β Transus temperature, these β grains transform to α -phase, according to the Burgers relation, in 12 possible transformation variants [81,[85][86][87]. Due to the high cooling rate, the probability of α formation is very high [37,88,89]. The rapid cooling leads to a martensitic transformation and a limitation of diffusional transformation [90]. Figure 3c shows the cross-section perpendicular to the BD. Areas with similar crystallographic orientations are observable inside the grain boundaries of the probable parent β grains. Figure 3d demonstrates the microstructure for post-treatment HT1. An unexpected, coarse lamella-like microstructures evolved instead of equiaxed grains [60]. The microstructure is dependent on the initial microstructure and dislocations before the heat treatment. HT2 results in a coarse microstructure and huge grains due to temperatures above β Transus and grain growth. The grain orientation seems random, see Figure 3e. Figure 3f exhibits the microstructure after stress relief treatment (HT3). Only minor changes in the microstructure occurred compared with the as-built condition. Areas with similar grain orientation are present and the prior β grain boundaries are noticeable. The detected phase distribution of α, α , and β titanium, respectively, is summarized in Table 1.
coarse microstructure and huge grains due to temperatures above βTransus and grai growth. The grain orientation seems random, see Figure 3e. Figure 3f exhibits th microstructure after stress relief treatment (HT3). Only minor changes in th microstructure occurred compared with the as-built condition. Areas with similar grai orientation are present and the prior β grain boundaries are noticeable. The detecte phase distribution of α, α′, and β titanium, respectively, is summarized in Table 1.   Figure 4b is the FFT pattern from the white circular area in Figure 4a. Figure 4c is the FFT pattern from the yellow circular area in Figure 4b. The diffraction pattern confirms the hexagonal crystal structure of α′ martensite in both laths. In addition, an HRTEM image of the interface between two laths (black square region in Figure 4a) is presented in Figure 4d. The red line highlights the interface between the two α′ martensite laths. Figure 4e is the magnified HRTEM image of the black square region in Figure 4d. The (01 11) plane of one lath intersects the ( 1 010) plane of another lath at 45°. Furthermore, Figure 4f shows the HAADF-STEM image and Figure 4g the EDS maps of the as-built sample. Here, laths of α′ martensite are apparent. The EDS maps of the black square region in Figure 4g show that no segregation of alloying elements is observable, and the titanium, aluminum, and niobium distribution is homogenous in the laths. Figure 5 shows the TEM images, HRTEM images, and diffraction pattern of the HT3 sample. The α laths are apparent in Figure 5a. In Figure 5a, some α laths are relatively   Figure 4b is the FFT pattern from the white circular area in Figure 4a. Figure 4c is the FFT pattern from the yellow circular area in Figure 4b. The diffraction pattern confirms the hexagonal crystal structure of α martensite in both laths. In addition, an HRTEM image of the interface between two laths (black square region in Figure 4a) is presented in Figure 4d. The red line highlights the interface between the two α martensite laths. Figure 4e is the magnified HRTEM image of the black square region in Figure 4d. The (0111) plane of one lath intersects the (1010) plane of another lath at 45 • . Furthermore, Figure 4f shows the HAADF-STEM image and Figure 4g the EDS maps of the as-built sample. Here, laths of α martensite are apparent. The EDS maps of the black square region in Figure 4g show that no segregation of alloying elements is observable, and the titanium, aluminum, and niobium distribution is homogenous in the laths.   Figure 5 shows the TEM images, HRTEM images, and diffraction pattern of the HT3 sample. The α laths are apparent in Figure 5a. In Figure 5a, some α laths are relatively coarse (≈170 nm). However, the α laths are relatively fine (≈ 40 nm). In this context, Figure 5b shows an HRTEM image taken from the region within the black square in Figure 5a. The lattice plane of two α laths is identified as (1010). The red lines point out the boundary between two α laths. For a deeper insight, Figure 5d is the magnified view of the black square region in Figure 5b. The interplanar spacing of the (1010) planes are measured as ≈ 0.252 nm. However, the two (1010) planes are misoriented by 18 • . The diffraction pattern in Figure 5c confirms the zone axis to be (1211). Again, the diffraction pattern shows the (1010) planes to be misoriented by 18 • . Figure 5e shows the HAADF-STEM image and EDS maps of the stress relief treated sample, where α laths are visible. Finally, Figure 5f is the EDS maps of a triple junction between three α laths from the black square region in      Figure 6 shows XRD patterns for the as-built and heat-treated Ti-6Al-7Nb samples. The XRD pattern of the sample in the as-built condition is characteristic for the hexagonal close-packed (hcp) α /α structure of titanium alloys. Due to the hexagonal structure of both the α and α phases with similar lattice parameters, a separation or distinction between both phases based on XRD is not possible [59]. Since additively manufactured alloys are characterized by high local cooling rates, the α martensitic phase is assumed as dominating component, based on the findings of Xu et al. [52]. Stress relief heat treatment at 600 • C does not lead to any significant microstructural changes in the sample HT3, since the α → α + β transformation is known to start at higher temperatures (≈760 • C) [59]. As shown for sample HT1, recrystallization annealing at 925 • C leads to the formation of the bcc β-structure of Ti and probably to a decomposition of α to α. HT2 treatment was carried out above the β Transus temperature and did result in decomposition of the as-fabricated α /α-phase [59]. Therefore, the β-phase is formed in the HT2 treated specimens. This is in good agreement with reported transformations and temperatures between 735 • C and 1050 • C for the α → β reaction [59]. Post-treatment HT1 and HT2 lead to a decrease in the α /α-Ti peak width, indicating grain growth. For the samples HT1 and HT2 additional peaks were observed between 2θ = 42-45 • . Due to the large width of the peaks, several similar phases with small differences in stoichiometry are most likely. As reported by Bolzoni et al., titanium and aluminum can form binary phases, e.g., titanium aluminates, above 660 • C due to diffusion processes [91]. Therefore, the formation of small fractions of Ti 3 Al, TiAl, and/or TiAl 3 is concluded for the heat treatments HT1 and HT2.
°C and 1050 °C for the α  β reaction [59]. Post-treatment HT1 and HT2 lead to a decrease in the α′/α-Ti peak width, indicating grain growth. For the samples HT1 and HT2 additional peaks were observed between 2θ = 42-45°. Due to the large width of the peaks, several similar phases with small differences in stoichiometry are most likely. As reported by Bolzoni et al., titanium and aluminum can form binary phases, e.g., titanium aluminates, above 660 °C due to diffusion processes [91]. Therefore, the formation of small fractions of Ti3Al, TiAl, and/or TiAl3 is concluded for the heat treatments HT1 and HT2.   Figure 7a. For the different conditions, the mean values of Young´s moduli, E, tensile yield strengths, Rp0,2, ultimate tensile strengths, Rm, and plastic elongations, A, are compared to the international standard for implants for surgery ISO 5832-11 in Figure 7c and Table 1 [92]. The martensitic structure of the additively processed (α + β) titanium leads to high values for the mechanical properties regarding ultimate tensile and yield strength compared with wrought and conventionally processed materials, especially with an α-dominant equiaxed microstructure [7]. The different material behavior might be explained by the different microstructures and phase composition, which have a strong impact on the mechanical properties. Typically, in conventional material, the size and morphology of the α grains determine the mechanical properties, while in LPBF the colony size and the size of the α or rather α′ laths control the properties. Plastic deformation tends to be the movement of dislocations. Conventional material with larger α grains enables the deformation with less dislocation pileups, whereas the smaller grains in the additively processed material increase the dislocation pileups [93]. The tensile properties are, inter alia, caused by the higher residual stresses, the martensitic transformation during fast cooling to the α′ instead of the α phase,    Figure 7a. For the different conditions, the mean values of Young s moduli, E, tensile yield strengths, R p0,2 , ultimate tensile strengths, R m , and plastic elongations, A, are compared to the international standard for implants for surgery ISO 5832-11 in Figure 7c and Table 1 [92]. The martensitic structure of the additively processed (α + β) titanium leads to high values for the mechanical properties regarding ultimate tensile and yield strength compared with wrought and conventionally processed materials, especially with an α-dominant equiaxed microstructure [7]. The different material behavior might be explained by the different microstructures and phase composition, which have a strong impact on the mechanical properties. Typically, in conventional material, the size and morphology of the α grains determine the mechanical properties, while in LPBF the colony size and the size of the α or rather α laths control the properties. Plastic deformation tends to be the movement of dislocations. Conventional material with larger α grains enables the deformation with less dislocation pileups, whereas the smaller grains in the additively processed material increase the dislocation pileups [93]. The tensile properties are, inter alia, caused by the higher residual stresses, the martensitic transformation during fast cooling to the α instead of the α phase, and grain refinement as a strengthening mechanism described by the Hall-Petch equation [94]. The Young s moduli vary within a certain range between 105 GPa-116 GPa. Due to heat treatments and microstructural changes, the average grain size increases. Sliding effects are mainly detected between grains; therefore, the breaking elongation should increase [95]. Another reason for changes in breaking elongation is the presence of β titanium. An increased amount of bcc β titanium should improve the ductile behavior [57,95]. EBSD phase analysis showed the highest fraction of β titanium for as-built conditions (3.0%), the lowest fraction for HT1 (1.7%), and between for HT2 (2.2%) and HT3 (2.3%), in accordance to the breaking elongation, see Figure 7c. According to the XRD analysis only for HT1 and HT2 β titanium is verified and may, therefore, explain the lower tensile strength for these conditions but not the reduced breaking elongations. and grain refinement as a strengthening mechanism described by the Hall-Petch equation [94]. The Young´s moduli vary within a certain range between 105 GPa-116 GPa. Due to heat treatments and microstructural changes, the average grain size increases. Sliding effects are mainly detected between grains; therefore, the breaking elongation should increase [95]. Another reason for changes in breaking elongation is the presence of β titanium. An increased amount of bcc β titanium should improve the ductile behavior [57,95]. EBSD phase analysis showed the highest fraction of β titanium for as-built conditions (3.0%), the lowest fraction for HT1 (1.7%), and between for HT2 (2.2%) and HT3 (2.3%), in accordance to the breaking elongation, see Figure 7c. According to the XRD analysis only for HT1 and HT2 β titanium is verified and may, therefore, explain the lower tensile strength for these conditions but not the reduced breaking elongations. In general, due to miniaturization of samples the fatigue and monotonic tensile behavior could be affected [42]. The monotonic material properties, such as Young´s moduli E, tensile yield strengths Rp0,2, ultimate tensile strengths Rm, plastic elongations A, the fatigue parameters-such as fatigue strength coefficients σ′f and exponents b-and fatigue ductility coefficients ε′f and exponents c, of the Coffin-Manson-Basquin equation, are given in Table 1.

Fracture Behavior
The fracture surface of the as-built Ti-6Al-7Nb alloy demonstrates mainly ductile fracture behavior, see Figure 7b. The propensity of cleavage fracture decreases with decreasing grain size. Therefore, the fracture surface of the additively processed materials shows cleavage facets with high amounts of dimples at the grain boundaries [58]. These fractured surfaces show small, shallow dimples on quasi-cleavage fracture surfaces, and transgranular facets, confirming the minor brittle fracture behavior of the additively processed specimens. Features, as a result of the LPBF fabrication, such as pores, unmelted powder particles, and defects occur on fracture surfaces in as-built and heat-treated conditions, see Figure 7d. These build defects typically are perpendicular to the build direction and pulled apart by the tensile load during monotonic and fatigue tests [99,100]. The fracture surfaces of fatigue tests tend to be smoother for lower strain amplitudes. Distinct fatigue striations, perpendicular to the crack growth direction, can be detected for all conditions and strain amplitudes, see Figure 8b. The striations are close to the probable crack initiation spot. Typical forced rupture areas are difficult to spot.

Characterization of Roughness, Cytocompatibility, and Antibacterial Effects
The roughness of the specimens without surface treatment (NT) exhibits values for Ra of 7.5 ± 0.3 µm and Rz of 49.6 ± 2.2 µm. Due to unmelted powder particles and the surface of the layer-wise fabrication, the roughness of NT specimens is higher than for the blasted specimens (ab) with Ra of 4.8 ± 0.2 µm and Rz of 34.3 ± 1.2 µm, which have smoother surfaces, on account of the mechanically post-processed treatment.
There was no effect of cell proliferation of the two examined samples on murine L-929 cells and human HUVEC cells. A slight tendency of diminished cell proliferation was observed in human HOS cells on blasted Ti-6Al-7Nb. Thus, the examined samples show a proper biocompatibility behavior for the different cell lines, used in the present study, see Figure 9. Furthermore, no increase in cytokine release of Il-6 was detected (data not shown). Concerning antibacterial effects, E. coli proliferation was not influenced by the  Figure 8a depicts the fatigue life of the different conditions. The fatigue performance for higher strain amplitudes (2N f < 10 2 reversals) of the as-built condition is better, compared with the HT1, HT2, and HT3 conditions. The fatigue performance in LCF range is connected to the monotonic tensile performance in terms of tensile strength. Higher strengths tend to lead to higher, tolerable strain amplitudes and higher fatigue life. Fatigue strength is generally high as higher monotonic strength hinders microplasticity and eventually local damage [42]. The curves of the as-built HT1 and HT3 conditions intersect around 10 2 cycles and are comparable up to 10 4 cycles. While the performance for higher strain amplitudes (2N f < 10 4 reversals) for the HT2 condition is the worst, the performance gets better for lower strain amplitudes. As the process-induced pores and defects are not affected by these post-treatments, other causes are likely to be decisive, such as the microstructures or residual stresses. For post treatment HT2 the reduced strength and the microstructural notches could lead to accumulation of local damage and finally results in early crack initiation [42]. The fatigue crack growth behavior of post-treated Ti-6Al-7Nb is affected by the microstructure. Depending on the crack growth direction and grain long axis different crack growth rates are probable [96][97][98]. For lower strain amplitudes, the three heat-treated conditions show superior fatigue behavior compared with the as-built conditions, attributed to the reduced residual stresses and microstructural features [52,[56][57][58][59]94]. Corresponding to the fatigue ductility exponent c, the heat-treated specimens show smaller gradients, while the as-built condition has the lowest value resulting in the shortest fatigue life. HT1 and HT3 exhibit close fatigue ductility exponent c and, therefore, are probably favorable for HCF applications.
In general, due to miniaturization of samples the fatigue and monotonic tensile behavior could be affected [42]. The monotonic material properties, such as Young s moduli E, tensile yield strengths R p0,2 , ultimate tensile strengths Rm, plastic elongations A, the fatigue parameters-such as fatigue strength coefficients σ f and exponents b-and fatigue ductility coefficients ε f and exponents c, of the Coffin-Manson-Basquin equation, are given in Table 1.

Fracture Behavior
The fracture surface of the as-built Ti-6Al-7Nb alloy demonstrates mainly ductile fracture behavior, see Figure 7b. The propensity of cleavage fracture decreases with decreasing grain size. Therefore, the fracture surface of the additively processed materials shows cleavage facets with high amounts of dimples at the grain boundaries [58]. These fractured surfaces show small, shallow dimples on quasi-cleavage fracture surfaces, and transgran-ular facets, confirming the minor brittle fracture behavior of the additively processed specimens. Features, as a result of the LPBF fabrication, such as pores, unmelted powder particles, and defects occur on fracture surfaces in as-built and heat-treated conditions, see Figure 7d. These build defects typically are perpendicular to the build direction and pulled apart by the tensile load during monotonic and fatigue tests [99,100]. The fracture surfaces of fatigue tests tend to be smoother for lower strain amplitudes. Distinct fatigue striations, perpendicular to the crack growth direction, can be detected for all conditions and strain amplitudes, see Figure 8b. The striations are close to the probable crack initiation spot. Typical forced rupture areas are difficult to spot.

Characterization of Roughness, Cytocompatibility, and Antibacterial Effects
The roughness of the specimens without surface treatment (NT) exhibits values for Ra of 7.5 ± 0.3 µm and Rz of 49.6 ± 2.2 µm. Due to unmelted powder particles and the surface of the layer-wise fabrication, the roughness of NT specimens is higher than for the blasted specimens (ab) with Ra of 4.8 ± 0.2 µm and Rz of 34.3 ± 1.2 µm, which have smoother surfaces, on account of the mechanically post-processed treatment.
There was no effect of cell proliferation of the two examined samples on murine L-929 cells and human HUVEC cells. A slight tendency of diminished cell proliferation was observed in human HOS cells on blasted Ti-6Al-7Nb. Thus, the examined samples show a proper biocompatibility behavior for the different cell lines, used in the present study, see Figure 9. Furthermore, no increase in cytokine release of Il-6 was detected (data not shown). Concerning antibacterial effects, E. coli proliferation was not influenced by the different samples, while the proliferation of S. aureus was reduced by the blasted surface of Ti-6Al-7Nb, see Figure 9. As described by Schweikl et al., blasting with Al2O3 particles led to incorporation of Al into the outer surface layers [101]. Therefore, the increased concentration of Al on the specimens' surfaces could result in decreasing the proliferation behavior of HOS cells and S. aureus [102][103][104]. Rough sandblasted surfaces with sharp ridges and edges could influence the proliferation and it seems to appear that the quality of cell contact on rough surfaces is related to the minimum width of the cavity [101]. As reported in various studies, blasting leads to the presence of severely plastic deformed layers and, therefore, to strain-hardening, as well as to compressive residual stresses [105][106][107]. Due to the antibacterial effect regarding the proliferation of S. aureus and because of probably better fatigue properties of blasted specimens, the specimens for the mechanical tests were blasted with Al2O3 particles.  As described by Schweikl et al., blasting with Al 2 O 3 particles led to incorporation of Al into the outer surface layers [101]. Therefore, the increased concentration of Al on the specimens' surfaces could result in decreasing the proliferation behavior of HOS cells and S. aureus [102][103][104]. Rough sandblasted surfaces with sharp ridges and edges could influence the proliferation and it seems to appear that the quality of cell contact on rough surfaces is related to the minimum width of the cavity [101]. As reported in various studies, blasting leads to the presence of severely plastic deformed layers and, therefore, to strain-hardening, as well as to compressive residual stresses [105][106][107]. Due to the antibacterial effect regarding the proliferation of S. aureus and because of probably better fatigue properties of blasted specimens, the specimens for the mechanical tests were blasted with Al 2 O 3 particles.

Conclusions
The microstructure and the monotonic and fatigue behaviors of additively processed Ti-6Al-7Nb were investigated. Different surface conditions were examined regarding biocompatible properties, such as cytocompatibility and antibacterial effects. The following conclusions can be drawn from these investigations:

•
Ti-6Al-7Nb shows significantly different microstructures in the as-built, stress-relieved, recrystallized, and β-annealed conditions. While the α /α structure is dominant within the as-built state, the formation and precipitation of the β-phase are manageable by a vacuum heat treatment above 925 • C as analyzed using XRD and EBSD.

•
There are significant differences in the monotonic tensile properties of the various conditions. Concerning the ISO values, the as-built and stress-relieved conditions (HT3) are favorable. Tensile and yield strength, as well as breaking elongation, are higher than the demanded values, but the specimens' dimensions have to be considered. HT1 and HT2 do not fulfill the requirements, probably due to the present β-phase. Heat treatments, such as HT1 and HT2, can significantly affect the microstructure and may tend to soften the lattice structures and decrease the residual stresses and, therefore, lead to significantly reduced tensile strength. • LCF life is higher for lower strain amplitudes in the heat-treated specimens than in as-built conditions, which can be attributed to decreased residual stresses as well as to microstructural differences. HT2 shows the highest fatigue life for lower strain amplitudes, while the as-built condition has higher service life for higher strain amplitudes. Regarding overall performance, quasi-static results, and LCF performance, stress relief treatment (HT3) is favorable. The material behavior for HCF loading still has to be determined. • Transgranular facets characterize the fracture surface of the additively processed Ti-6Al-7Nb, confirming the ductile behavior. Small, shallow dimples on quasi-cleavage fracture surfaces are visible. Fatigue fracture surfaces are characterized by fatigue striations and remaining forced rupture surfaces. Defects may have less impact on quasi-static but a high impact on fatigue behavior. • Both surfaces, the untreated and the blasted, show good biocompatibility in different cell types (fibroblasts, osteosarcoma cells, and endothelial cells). Only a slight antiproliferative effect was observed for blasted Ti-6Al-7Nb in osteoblasts. An increase in cytokine release of Il-6 was not observed.

•
Blasting with high-grade Al 2 O 3 is preferable regarding biocompatibility and antibacterial effects. Blasted Ti-6Al-7Nb exhibits an antibacterial effect against S. aureus in comparison with not post-treated Ti-6Al-7Nb. E. coli was able to grow on both surfaces of Ti-6Al-7Nb similarly. On account of remaining aluminum on the blasted surface, glass bead blasting could be taken into consideration.

Data Availability Statement:
The data that support the findings of this study are available from the corresponding author upon reasonable request.