1 Introduction

The major requirement for metallic materials in biomedical applications includes good mechanical properties, i.e., sufficient strength and ductility, low Young’s modulus, high wear resistance and fatigue strength, as well as corrosion resistance and biocompatibility (no adverse effect to a human body).[1,2,3] Among metallic materials used in biomedical applications, titanium and its alloys are attractive due to their relatively low Young’s modulus (compared to stainless steels and Co alloys), high-strength/weight ratio, superior corrosion resistance, and excellent biocompatibility.[4,5,6] Furthermore, titanium and its alloys are better candidates for load-bearing biomedical implants than polymers and ceramics due to their high mechanical strength and fracture toughness.[7]

The α+β-type Ti-6Al-4V alloy, developed for aerospace industry, has been used in biomedical applications for a long time.[8,9] More recently, Ti-6Al-7Nb and Ti-5Al-2.5Fe were introduced into the market to avoid the adverse effect of vanadium (cytotoxic effects and adverse tissue reaction[4]). However, it has been demonstrated that Al also causes long-term neurological diseases such as Alzheimer’s, osteomalacia, and peripheral neuropathy.[1,4,10] Therefore, Al- and V-free alloys gained more attention in biomedical applications. Commercially pure titanium (CP-Ti) has also been placed in the biomedical industry, since it has better chemical biocompatibility than that of Ti-6Al-4V and Ti-5Al-2.5Fe.

Strength is one of the most important requirements for load-bearing biomaterials. The tensile strength of α-type CP-Ti (grade 2) is about 350 MPa, while that of α+β alloys (Ti-6Al-6V, Ti-6Al-7Nb, and Ti-5Al-2.5Fe) varies in the range of 900 to 1050 MPa.[9] However, the major drawback of these materials is high Young’s modulus ranging between 100 and 120 GPa,[9] which is well above Young’s modulus of bone (10 to 30 GPa).[11] This stiffness mismatch between implant and bone causes a stress shielding effect, which may result in loosening of the implant or deformation of the bone.[3,11,12,13] Thus, besides high strength, the load-bearing biomaterial has to possess relatively low Young’s modulus in order to avoid the stress shielding effect. Table I shows the comparison of mechanical properties of typical Ti alloys.

Table I Mechanical Properties of Various Types of Ti Alloys
Full size table

Recently, β-type titanium alloys have attracted significant attention, because their alloying β-stabilizing elements, such as Nb, Zr, Ta, and Mo, are biocompatible and nontoxic.[1,4,14] In addition, compared to α and α+β alloys, they exhibit lower Young’s modulus, which is closer to that of bone and has better corrosion resistance.[15,16] The Ti-13Nb-13Zr (TNZ) ternary alloy is one of the most promising near β-type titanium alloys for biomedical applications, developed by Davidson and Kovacs in the 1990s.[14] In this alloy, Nb is a β-phase stabilizer, which also reduces Young’s modulus, while Zr is a neutral element since it is isomorphous with both α and β phases.[13,17] The Young’s modulus of TNZ is about 79 GPa, which is lower than conventional titanium alloys, Co-Cr alloys, and stainless steel (Table I). Also, its corrosion resistance and biocompatibility are better due to the Zr addition, which has high hemocompatibility.[18,19] Thus, TNZ has superior properties to those of conventional load-bearing α and α+β biomaterials. However, its strength is lower (Table I), which implies a necessity of a marked enhancement in its mechanical strength.

One of the efficient ways to improve the mechanical strength is to apply severe plastic deformation (SPD) methods to decrease the grain size to the ultrafine-grained (UFG) range (i.e., below 1 μm) in accordance with the Hall–Petch relationship.[20,21,22,23,24] To effectively reduce the grain size, SPD methods, such as accumulative roll bonding,[25] high pressure torsion (HPT),[26] and equal channel angular pressing,[27] can be used. Besides these typical SPD methods, hydrostatic extrusion (HE) is also an efficient way of grain refinement. The main difference of HE compared to conventional extrusion is the existence of a fluid (hydrostatic medium) between the container wall and billet, thus reducing friction. However, HE is a fast process and adiabatic heating phenomena are likely to take place. Moreover, hydrostatic pressure allows processing even hard-to-deform materials, as it hinders crack formation and propagation.[28] Hence, HE is an attractive method to produce bulk UFG materials. The previous studies demonstrated that its efficiency in terms of refinement depends on the material features and processing conditions.[29,30,31] In this study, a near β-type ternary TNZ alloy was subjected to the HE process at room temperature (RT) in two stages, and its effect on mechanical properties, microstructure, and phase transformation behavior was investigated with an aim of producing a high-strength, low-modulus, biocompatible titanium-based implant material.

2 Experimental

The material used in this study is a near β-type TNZ ternary alloy intended for biomedical purposes (ASTM F1713) supplied as a rod with a diameter of 30 mm. Its chemical composition is given in Table II.

Table II Chemical Composition of TNZ Alloy (Weight Percent)
Full size table

The HE process was carried out at RT in two stages without intermediate annealing. The material diameter was reduced to 20 mm (denoted as HE20) by the first stage and to 16 mm by the second stage (denoted as HE16), which correspond to a total true strain of 2.49 (calculated according to ε = 2 ln(d i/d f), where d i is initial diameter and d f is final diameter) and total area reduction of 3.81 (calculated according to (d i/d f)2). For each individual stage, the hydrostatic pressure measured was about 1 GPa (Table III). The billet was water cooled at the die exit to limit the effect of adiabatic heating. Further, HE20 and HE16 samples were donated as HE20T and HE16L, where T stands for the transverse section, which is perpendicular to the extrusion direction, and L stands for the longitudinal section, which is parallel to the extrusion direction.

Table III HE Parameters of Individual Stages of Deformation
Full size table

Dog-bone-shaped tensile test samples (five samples for each condition with 4-mm diameter and 20-mm length) were prepared along the extrusion direction to investigate the tensile properties and Young’s modulus values using a Zwick/Roell Z250 testing machine. The tensile tests were performed at RT with a strain rate of 8 × 10−3 s−1.

Vickers microhardness measurements and tensile tests were carried out to evaluate the mechanical properties of the material. Microhardness was determined by a Zwick/Roell ZHU 2.5 testing machine with a 4.9 N of load (HV0.5) for a 12-second dwell time with a step size of d/10 (d stands for the diameter of the samples), and 10 indents were applied to obtain the average value.

For the measurements of X-ray diffraction (XRD) patterns, an AXS Bruker D8 diffractometer equipped with a GAADS area detector was used. Cu K α radiation with a wavelength λ Cu-Kα = 0.15406 nm was applied while the spot size amounted to 0.8 mm. To evaluate the phase fractions, corresponding XRD patterns were subjected to a Rietveld-type refinement procedure using the X’pert High score plus software package. Refinement parameters, such as expected residuals (R exp), weighted residuals (R wp), and goodness of fit (GOF), were evaluated. For texture measurements, the distribution of diffraction intensities from four crystallographic planes, {10-10}, {0002}, {10-11}, and {10-12}, were measured on the center of the sample in a way that the sample normal became parallel to the extrusion axis. The texture results were represented by pole figures and generated through Labotex software[32] using a c/a ratio of 1.58 for the current material.

Initial samples were mechanically ground with silicon carbide paper and polished by 1-μm diamond paste and then etched by Kroll’s reagent (3HF:6NO3:100H2O) to reveal the microstructure by light microscopy (LM). The microstructure of HE-processed samples was investigated by a JEOL JEM 1200EX transmission electron microscope (TEM) at an accelerating voltage of 120 kV. Disc-shaped TEM specimens with a diameter of 3 mm and a thickness of 100 μm were prepared by mechanical grinding and then electropolished by a Struers Tenupol-5 with a Struers A3-I electrolyte at 32 V operating voltage at (278 K) 5 °C.

3 Results and Discussion

3.1 Mechanical Properties

Figure 1 shows the billets after the first and second stages of the HE process. It should be noted that the surface of the billets has no macroscopic defects, which demonstrates the smooth flow of TNZ alloy during HE.

Fig. 1
figure 1

Appearance of TNZ samples after HE

Full size image

The engineering stress-strain curves of initial and extruded samples obtained in tensile tests are presented in Figure 2. Both extruded samples exhibit enhanced yield and ultimate tensile strength. After the first stage of HE with a true strain of 1.32, the yield strength (σ y ) increased by about 40 pct from 690 to 960 MPa and the ultimate tensile strength (σ UTS) increased from 845 to 1100 MPa. Subsequent second stage with a true strain of 1.17 did not cause significant changes in mechanical properties: both σ y and σUTS values only slightly increased to 975 and 1200 MPa, respectively (Table IV). This behavior is typical of SPD processing, where major changes take place for lower accumulated strains and then tend to saturate.

Fig. 2
figure 2

Tensile curves of initial and HE-processed samples

Full size image
Table IV Mechanical Properties and Phase Volume Fractions of Initial and HE-Processed Samples
Full size table

It is known that mechanical strengthening of a material by SPD is due to the grain size refinement and high density of accumulated defects induced by applied strain.[27,33] It also markedly depends on the presence of existing phases, since each crystallographic lattice has a specific deformation mechanism. For instance, it has been reported that the strength of the α phase (hexagonal) is almost 3 times higher than the β phase (bcc).[34,35] Our previous study on binary Ti-45Nb β alloy[36] revealed that HE processing with a total true strain of 3.5 brings about an increase in σ UTS of about 50 pct (from 445 to 663 MPa). Another study on α-type CP-Ti revealed a 100 pct increase in σ UTS (from 583 to 1148 MPa) after five stages of HE with a total true strain of 2.77.[37] The relative values of the strength increase for α-type CP-Ti are higher than those of near β-type TNZ and β-type Ti-45Nb.

Despite enhanced strength, elongation to failure (ε) has been reduced from 17.5 to 9.8 pct by the first stage and to 8.8 pct by the second stage of HE (Table IV). However, such values of elongation indicate that extruded TNZ alloy is still capable of accommodating further plastic deformation. This loss of ductility is typical of SPD-processed materials due to hardening of the material.[22] However, it should be noted that despite a decrease of ductility, it is still comparable to that of the Al and V containing titanium alloys such as Ti-6Al-4V.

A similar trend as for tensile strength was observed for microhardness changes—the microhardness increased significantly after the first and only slightly after the second extrusion pass. Initial microhardness of 271 HV0.5 measured on the transverse section increased to 300 and 305 HV0.5 for HE20T and HE16T, respectively. Compared to the transverse section, microhardness values of the longitudinal section are higher by about 7 pct, i.e., 320 and 323 HV0.5 for HE20L and HE16L, respectively (Table IV). This is clearly seen in the work-hardening graph illustrated in Figure 3 and can be attributed to the anisotropy of defect density, crystallographic orientation, and grain size distribution. The effect of the anisotropy after HE was investigated in detail in the previous work of the authors[36] and by Moreno-Valle et al.[38]

Fig. 3
figure 3

Microhardness vs true strain of initial and HE-processed samples

Full size image

The values of Young’s modulus determined by tensile tests are summarized in Table IV. It was revealed that HE processing of TNZ leads to a decrease in Young’s modulus from 78.2 for initial to 73.2 GPa for HE20 then to 69.6 GPa for the HE16 sample. The same effect of decreasing the Young’s modulus from 64 to 60 GPa by increasing strain was observed by Yilmazer et al. for Ti-29Nb-13Ta-4.6Zr β-type titanium alloy after the HPT process and attributed to increased volume fraction of nonequilibrium boundaries and triple junctions.[39] In order to explain the changes in the mechanical properties caused by HE processing, detailed phase and texture analyses as well as microstructure characterizations were carried out. In the case of Young’s modulus, it is known that the major factors influencing its value include chemical composition, phase composition, and texture.

3.2  Phase Transformation Analysis by XRD

Figure 4 shows the XRD results obtained from transverse and longitudinal sections (perpendicular and parallel to the extrusion direction, respectively) of the samples for all conditions from their center regions. Initial and extruded sample peaks correspond to α′ and β lattice structures for both transverse and longitudinal sections. Furthermore, in the case of the HE20T sample, the intensities for basal (0002) and pyramidal (10-11) planes are significantly higher than that of initial (Figure 4(a)). In the case of the HE16T sample, the basal plane intensity is hardly affected and a slight decrease of the α(10-11) plane intensity is observed. For the samples taken from longitudinal sections, both the basal and the α(10-11) plane show almost the same intensity for HE20L; however, the intensity of the basal plane increases for HE16L (Figure 4(b)).

Fig. 4
figure 4

XRD results of initial and HE-processed samples: (a) initial and transverse sections and (b) longitudinal section

Full size image

Peak broadening and shifting phenomena of initial vs HE16T and HE16L samples are shown in Figure 5. Broadening of the peaks is due to small crystallite size and increased dislocation density along with internal microstress, and the peak shifting is related to planar faults and internal stress.[40]

Fig. 5
figure 5

Peak broadening and shifting in XRD patterns: (a) transverse section and (b) longitudinal section

Full size image

The Rietveld refinement procedure was used to estimate the phase fractions of α′ and β phases (given in weight percent) for the initial and the extruded samples, as shown in Table IV. The initial sample contains 71 pct of α′ martensite phase and 29 pct of β phase (R exp = 6.57, R wp = 12.9, and GOF = 3.87), while the HE16T sample includes 77 pct of α′ martensite phase and 23 pct of β phase (R exp = 6.46, R wp = 12.46, and GOF = 3.71), approximately. Such changes in phase composition may affect both mechanical strength and Young’s modulus. The increased fraction of α′ martensite phase causes the strength to increase, while sufficient ductility is ascribed to the remaining β phase.[21] Also, the Young’s modulus is expected to increase due to the higher phase fraction of α′ martensite phase. Since this is not the case observed, an additional factor may play a role in describing the decrease of the Young’s modulus. Therefore, a texture study has been implemented.

The experimental (0001) pole figures for HE20T and HE16T in the extrusion system are illustrated in Figure 6. For HE20T (true strain of 1.32), {0001} basal fiber texture, in addition to relatively weak {10-10} 〈0001〉 and {10-10} 〈-2110〉 texture components, is observed (Figure 6(a)). With the increase of strain to 2.49 (in the case of HE16T), the texture persists; however, the basal texture becomes less prominent and the strength of other texture components dominates (Figure 6(b)).

Fig. 6
figure 6

(0001) and (10-10) pole figures of HE-processed samples: (a) HE20T (max. 3.2 mrd) and (b) HE16T (maximum 3 mrd) in the extrusion system

Full size image

The decrease in the Young’s modulus of TNZ after the HE process can be explained by means of the texture evolution, which leads to anisotropic behavior of the material. In hexagonal titanium, the Young’s modulus exhibits a maximum on the (0001) basal plane and minimum on the (10-10) prismatic plane.[41] Due to the fact that in the HE16Tsample the basal fiber texture is less prominent than that of HE20T, the Young’s modulus decreases slightly.

3.3  Microstructure Analysis

The microstructure of the initial sample observed by LM after polishing and etching by Kroll’s etchant is shown in Figure 7. The microstructure shows acicular martensitic α′ laths in various directions embedded in the β matrix. Detailed investigations made by TEM for the initial sample (shown in Figure 8(a)) show that the α′ laths have width and length in the range of 50 to 300 nm and 300 nm to a few microns, respectively. It is also clearly seen in Figure 8(a) that α′ laths are almost free of dislocations and corresponding spot patterns indicate coarse-grained initial microstructure.

Fig. 7
figure 7

Microstructure of initial sample taken by light microscope

Full size image
Fig. 8
figure 8

TEM bright-field images with SADP: (a) initial and (b) HE16T

Full size image

After the HE process, significant refinement of α′ laths occurs, as illustrated in Figure 8(b). They are elongated and bent. Elongated α′ laths with kinking morphology are expected from the flow behavior of the HE. This morphology has a strong effect on the texture since the rotation of the α′ laths is easier.[42] Besides refinement of α′ laths, applied strain introduced a high density of dislocations. All these microstructural changes are responsible for the observed significant increase in mechanical strength, i.e., smaller grain size increases the strength according to the Hall–Petch relationship, and increased dislocation density causes dislocation hardening.[20,23]

Figure 9 shows TEM micrographs of the HE16L sample. The bright-field image shows α′ laths elongated within the extrusion direction, which is typical of material deformed by the HE process. One can notice that neighboring laths often vary in contrast, which suggests different orientations between them. At the same time, contrast within the α′ laths observed in the direction perpendicular to boundaries is also changing. This can be a result of dislocation substructure inside of them that makes it possible to distinguish slight deviations from overall orientations of observed laths. Additionally, the effect of internal strains may amplify this effect.

Fig. 9
figure 9

HE16L specimen: (a) bright field, (b) corresponding diffraction pattern, (c) dark field taken using α′-Ti spot, and (d) dark field taken with β-Ti spot

Full size image

Analysis of the diffraction pattern presented in Figure 9(b) indicates that two phases, hexagonal α′ and β, are present since two low index rings are visible, each attributed to different phases. Dark-field images, taken using (0001) α and (011) β spots from α and β phases, allow revealing phase distribution in the microstructure (Figures 9(c) and (d)). The different phases are alternately stacked and form the layered structure. Similar observations were performed for HE16T and HE16L samples. Figure 10(a) presents the transverse section of HE16, and α′ lath elongation is clearly noticeable. The diffraction pattern of the same area shown in Figure 10(b) and well-developed rings with the same scattering planes as in the longitudinal section were observed. The ring form of diffraction pattern corresponding to HE16T suggests high misorientation spread of both phases.

Fig. 10
figure 10

HE16T specimen: (a) TEM bright-field image and (b) corresponding diffraction pattern

Full size image

The mechanism of formation of such a structure is not known at this stage and requires more deep analysis. However, the authors believe that it is connected with phase transformation occurring during HE. Nevertheless, investigations performed on two sections of HE-processed TNZ suggest this deformation structure arises from stacking of laths out of alternating α′ martensite and β phases, with thickness below 100 nm.

4 Conclusions

The near β-type TNZ alloy was subjected to HE in two passes with a total true strain of 2.49, and its effect on microstructure, mechanical properties, and phase transformation was studied. Significant strengthening was achieved (σ UTS from 845 to 1200 MPa) due to grain size refinement and increased storage of dislocations within the grain interiors. Elongation to fracture still amounts to 8.8 pct after HE processing. The increase of microhardness was more effective in longitudinal sections (from 261 to 323 HV0.5) than transverse sections (from 271 to 305 HV0.5) as a result of anisotropic plastic deformation. The Young’s modulus showed about 10 pct lower value after HE processing due to pronounced {0001} basal texture. The mechanical properties found after HE suggest the material for use in biomedical applications.