Influence of Boron Additions and Heat Treatments on the Fatigue Resistance of CoCrMo Alloys

Cobalt-based alloys are widely used in the manufacture of joint prostheses. In this study, the effect of boron additions and heat treatment on the ASTM F75 was evaluated by rotating bending fatigue. The boron ranged from 0.06–1 wt %. The alloys were tested in as-cast and heat-treated conditions. In the as-cast condition, the infinite life was observed at 380 MPa, improving to 433–615 MPa according to the amount of boron added. In the heat treatment condition, the fatigue resistance was improved only in the base alloy. The addition of 0.06 wt % boron and heat treatment led to the same resistance as in the as-cast condition. Adding large amounts of boron combined with heat treatment diminished the fatigue limit. The fracture analysis revealed primarily brittle behaviour with some ductile features even on the same sample; only the heat-treated alloy with 0.06 wt % boron was clearly ductile. This alloy also exhibited notably better toughness to crack propagation.


Introduction
Nowadays, the standard treatment for degenerative joints is to replace them with medical devices, such as knee and hip prostheses, 70-80% of which are made of metallic biomaterials [1]. Such devices face severe environmental conditions. The concentrations of chloride ions in the serum and the interstitial fluid are 113 and 117 mEq L −1 , respectively, while the oxygen dissolved in blood is one-fourth of that in air [2]. Also, the pH in body fluids decreases to 5.2 when a material is implanted in the hard tissue and takes about two weeks to recover to 7.4 [3]. In addition, metallic biomaterials have to fulfil strict requirements. The main metallic biomaterials are stainless steels, cobalt-chrome alloys, and titanium alloys, each with their own advantages and disadvantages. Titanium alloys present a lower elasticity modulus compared to stainless steel and Co-Cr alloys, which reduce the stress shielding effect [4,5]. The effect of different alloying elements on the microstructure of titanium alloys and their elastic modulus has been reported [6][7][8]. However, titanium implants are fatigue damaged due to wearing and fretting, which leads to corrosion pits on the surface and consequently a decrease in fatigue resistance [4]. Cobalt alloys exhibit high corrosion resistance and excellent wear resistance [1]. The clinical use of these alloys for long periods of time has shown good biocompatibility in bulk form [2]. However, the fatigue process assisted by corrosion has been reported as the failure  Table 1). The specimens were tested in the as-cast and after heat treatment conditions. The heat treatment was performed at 1200 • C for 1 h followed by water quenching. The temperature for the heat treatment was chosen from the results observed in previous works [21,22]. Table 2 identifies the various alloys.

Specimens and Equipment
Fatigue samples were machined at the dimensions required for evaluation in the rotating bending fatigue equipment described in a previous work [23]. The loading was applied by constant deflection, implying that the load decreased as the crack grew. Figure 1 shows the geometry and dimensions of the fatigue specimens. The specimens were tested in the as-cast and after heat treatment conditions. The heat treatment was performed at 1,200°C for 1 h followed by water quenching. The temperature for the heat treatment was chosen from the results observed in previous works [21,22]. Table 2 identifies the various alloys.

Specimens and Equipment
Fatigue samples were machined at the dimensions required for evaluation in the rotating bending fatigue equipment described in a previous work [23]. The loading was applied by constant deflection, implying that the load decreased as the crack grew. Figure 1 shows the geometry and dimensions of the fatigue specimens.

Stress and Stress Concentration Factor
The evaluated fatigue samples had a stress concentration to promote the fracture in a controlled zone. The effect of a notch on the local stress depends on the geometry and root radius; for this specific part in bending, the stress concentration factor (Kt) was determined as follows [24]: where σnom is the nominal stress and σmax is the maximum local elastic stress acting at the notch. The nominal stress at the surface of a bending shaft is calculated as [24]: where M is the bending moment and d is the smaller diameter of the sample. The σmax was determined by finite element analysis (FEA) in the ANSYS Workbench package (version 13.0, USA) with 3,210 nodes and 1,794 elements; the estimated stresses are shown in Figure 2. Using the σmax obtained via FEA and the σnom computed using Equation (2), a Kt = 5.4 was obtained from Equation (1). Thus, the reported stresses are the values of σmax calculated using this Kt value.

Stress and Stress Concentration Factor
The evaluated fatigue samples had a stress concentration to promote the fracture in a controlled zone. The effect of a notch on the local stress depends on the geometry and root radius; for this specific part in bending, the stress concentration factor (K t ) was determined as follows [24]: where σ nom is the nominal stress and σ max is the maximum local elastic stress acting at the notch. The nominal stress at the surface of a bending shaft is calculated as [24]: where M is the bending moment and d is the smaller diameter of the sample. The

Experimental Procedure
The fatigue tests were performed in cantilever rotating bending with constant deflection at a frequency of 20 Hz and a stress ratio of R = −1. To construct the S-N curves (stress-number of cycles), the experiments were repeated three times at each stress level for each material condition. It was considered an infinite life at 10 7 cycles. During testing, the load data were acquired to consider how the load decreases with macroscopic crack growth. Fracture surface analysis was carried out by means of scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS) in a JEOL JSM-6510LV (Tokyo, Japan).

S-N Curves
The S-N curves for the as-cast and heat-treated conditions are shown in Figure 3. A noticeable scatter of results is observed for the same alloy at the same stress level due to factors such as the microstructure and embedding of particles from the ceramic mould in the samples. Less dispersion could be expected if the manufacturing process were controlled more carefully.

Experimental Procedure
The fatigue tests were performed in cantilever rotating bending with constant deflection at a frequency of 20 Hz and a stress ratio of R = −1. To construct the S-N curves (stress-number of cycles), the experiments were repeated three times at each stress level for each material condition. It was considered an infinite life at 10 7 cycles. During testing, the load data were acquired to consider how the load decreases with macroscopic crack growth. Fracture surface analysis was carried out by means of scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS) in a JEOL JSM-6510LV (Tokyo, Japan).

S-N Curves
The S-N curves for the as-cast and heat-treated conditions are shown in Figure 3. A noticeable scatter of results is observed for the same alloy at the same stress level due to factors such as the microstructure and embedding of particles from the ceramic mould in the samples. Less dispersion could be expected if the manufacturing process were controlled more carefully.

Experimental Procedure
The fatigue tests were performed in cantilever rotating bending with constant deflection at a frequency of 20 Hz and a stress ratio of R = −1. To construct the S-N curves (stress-number of cycles), the experiments were repeated three times at each stress level for each material condition. It was considered an infinite life at 10 7 cycles. During testing, the load data were acquired to consider how the load decreases with macroscopic crack growth. Fracture surface analysis was carried out by means of scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS) in a JEOL JSM-6510LV (Tokyo, Japan).

S-N Curves
The S-N curves for the as-cast and heat-treated conditions are shown in Figure 3. A noticeable scatter of results is observed for the same alloy at the same stress level due to factors such as the microstructure and embedding of particles from the ceramic mould in the samples. Less dispersion could be expected if the manufacturing process were controlled more carefully. The as-cast condition showed an improvement in the fatigue resistance with the addition of boron. The base alloy in this condition showed an infinite fatigue life at 380 MPa, whereas the alloys modified with boron exhibited increased resistance of 433, 487, 541 and 615 MPa according to the 0.06, 0.25, 0.5 and 1 wt % B content, respectively. The maximum fatigue resistance was observed at 615 MPa in the alloy with 1 wt % B, which is higher than the 300 MPa reported previously for the ascast condition [11,19]. Sudhakar and Wang [25] reported an endurance limit of 387 MPa for a Co alloy with Ni additions. However, the use of Ni is limited in medical applications because of allergic reactions; ASTM F75-07 alloy restricts the Ni content to 0.5 wt % [26].
In contrast, among the heat-treated alloys, only the base alloy exhibited improved fatigue resistance compared with the as-cast condition, recording a value of 445 MPa. The fatigue limit for the 0.06B-HT alloy was 433 MPa, the same as that in the as-cast condition. The 0.25%B-HT and 0.5%B-HT alloys exhibited a fatigue resistance of 314 and 270 MPa, respectively. Regarding the 1%B-HT alloy, one sample withstood 9.85 million cycles at 615 MPa of stress amplitude.
The load data recorded during the fatigue tests allowed an estimation of the number of cycles occurring between the load drop and final fracture, which is related to the propagation rate of macroscopic cracks. Table 3 presents representative examples for each material evaluated regarding the approximate number of elapsed cycles in the final stage of the tests. In this case, the 0.06B-HT alloy displayed the best performance by enduring significantly more cycles than any other alloy before completing the failure, whereas the 1B-HT alloy exhibited the fastest crack propagation. The higher resistance to crack growth of the 0.06B-HT alloy could be due to the combined effect of (i) reducing microstructural casting defects by adding a small amount of B, as reported by Zhuang, with trace elements [17] and (ii) removing the detrimental phases by heat treatment, as described by Dobbs and Robertson [19]. In contrast, adding larger amounts of B increased the precipitation of boron carbides, leading to the alloys being more brittle and cracks usually propagating faster since the materials were more brittle.  The as-cast condition showed an improvement in the fatigue resistance with the addition of boron. The base alloy in this condition showed an infinite fatigue life at 380 MPa, whereas the alloys modified with boron exhibited increased resistance of 433, 487, 541 and 615 MPa according to the 0.06, 0.25, 0.5 and 1 wt % B content, respectively. The maximum fatigue resistance was observed at 615 MPa in the alloy with 1 wt % B, which is higher than the 300 MPa reported previously for the as-cast condition [11,19]. Sudhakar and Wang [25] reported an endurance limit of 387 MPa for a Co alloy with Ni additions. However, the use of Ni is limited in medical applications because of allergic reactions; ASTM F75-07 alloy restricts the Ni content to 0.5 wt % [26].
In contrast, among the heat-treated alloys, only the base alloy exhibited improved fatigue resistance compared with the as-cast condition, recording a value of 445 MPa. The fatigue limit for the 0.06B-HT alloy was 433 MPa, the same as that in the as-cast condition. The 0.25%B-HT and 0.5%B-HT alloys exhibited a fatigue resistance of 314 and 270 MPa, respectively. Regarding the 1%B-HT alloy, one sample withstood 9.85 million cycles at 615 MPa of stress amplitude.
The load data recorded during the fatigue tests allowed an estimation of the number of cycles occurring between the load drop and final fracture, which is related to the propagation rate of macroscopic cracks. Table 3 presents representative examples for each material evaluated regarding the approximate number of elapsed cycles in the final stage of the tests. In this case, the 0.06B-HT alloy displayed the best performance by enduring significantly more cycles than any other alloy before completing the failure, whereas the 1B-HT alloy exhibited the fastest crack propagation. The higher resistance to crack growth of the 0.06B-HT alloy could be due to the combined effect of (i) reducing microstructural casting defects by adding a small amount of B, as reported by Zhuang, with trace elements [17] and (ii) removing the detrimental phases by heat treatment, as described by Dobbs and Robertson [19]. In contrast, adding larger amounts of B increased the precipitation of boron carbides, leading to the alloys being more brittle and cracks usually propagating faster since the materials were more brittle.

Fatigue Crack Surfaces
Analysing the fracture surfaces revealed mainly brittle behaviour; however, ductile features were also observed on the same sample. This could possibly be due to an increase in the strain-induced HCP phase after cyclic loading and a small amount of plastic strain after fracture due to the phase transformation [12]. Gueler [12] also stated that the amount of phase transformation was smaller with higher carbon content and smaller grain size. Figure 4 shows the brittle fracture surfaces in alloys without B, along with facets and striations. The fracture observed for the 0B-HT condition exhibited a high concentration of non-metallic inclusions from the ceramic mould, which act as crack generators.

Fatigue Crack Surfaces
Analysing the fracture surfaces revealed mainly brittle behaviour; however, ductile features were also observed on the same sample. This could possibly be due to an increase in the straininduced HCP phase after cyclic loading and a small amount of plastic strain after fracture due to the phase transformation [12]. Gueler [12] also stated that the amount of phase transformation was smaller with higher carbon content and smaller grain size. Figure 4 shows the brittle fracture surfaces in alloys without B, along with facets and striations. The fracture observed for the 0B-HT condition exhibited a high concentration of non-metallic inclusions from the ceramic mould, which act as crack generators.  Figure 5 shows the fracture surfaces of alloys with 0.06 wt % B. The 0.06B-AC alloy exhibited a transgranular fracture with microcracks and smoother reliefs, whereas the 0.06B-HT alloy exhibited a more ductile fracture with plastic deformation and dimples. The highest resistance to crack growth was also observed under this material condition. In a previous study, this alloy also showed good corrosion resistance [20]. The dark areas in some overview pictures were coloured with ink to identify certain areas during observations using scanning electron microscopy.   Figure 5 shows the fracture surfaces of alloys with 0.06 wt % B. The 0.06B-AC alloy exhibited a transgranular fracture with microcracks and smoother reliefs, whereas the 0.06B-HT alloy exhibited a more ductile fracture with plastic deformation and dimples. The highest resistance to crack growth was also observed under this material condition. In a previous study, this alloy also showed good corrosion resistance [20]. The dark areas in some overview pictures were coloured with ink to identify certain areas during observations using scanning electron microscopy.

Fatigue Crack Surfaces
Analysing the fracture surfaces revealed mainly brittle behaviour; however, ductile features were also observed on the same sample. This could possibly be due to an increase in the straininduced HCP phase after cyclic loading and a small amount of plastic strain after fracture due to the phase transformation [12]. Gueler [12] also stated that the amount of phase transformation was smaller with higher carbon content and smaller grain size. Figure 4 shows the brittle fracture surfaces in alloys without B, along with facets and striations. The fracture observed for the 0B-HT condition exhibited a high concentration of non-metallic inclusions from the ceramic mould, which act as crack generators.  Figure 5 shows the fracture surfaces of alloys with 0.06 wt % B. The 0.06B-AC alloy exhibited a transgranular fracture with microcracks and smoother reliefs, whereas the 0.06B-HT alloy exhibited a more ductile fracture with plastic deformation and dimples. The highest resistance to crack growth was also observed under this material condition. In a previous study, this alloy also showed good corrosion resistance [20]. The dark areas in some overview pictures were coloured with ink to identify certain areas during observations using scanning electron microscopy.    Figure 6 shows the fracture surfaces of alloys with 0.25 wt % B. The 0.25B-AC alloy exhibited a transgranular crack at the beginning that propagated in a radial direction, with dimples in the propagation zone and an intergranular final fracture. The 0.25B-HT alloy exhibited zones of faceted 'stair-like' deformation (brittle) and zones with dimples and microvoids (ductile). The fracture surfaces of alloys with 0.5 wt % B are shown in Figure 7. The 0.5B-AC alloy exhibited brittle striations that originated from the martensitic transformation induced by cyclic deformation at high crack growth [12,17]. The 0.5-HT alloy had an intergranular appearance at crack onset and during propagation and a transgranular appearance at the end. The fracture surfaces in alloys with 1 wt % B are shown in Figure 8. The analysed sample of 1B-AC alloy exhibited a transgranular crack at the beginning and in the propagation zone and intergranular appearance at the end; the crack nucleated on a boron carbide. The 1B-HT alloy was the most brittle among the studied alloys; a stair-like fracture surface is observed with pyramids and cleavages. The pyramids arose from the intersection of three planes in a secondary cleavage crack [27]. The fracture in this 1B-HT sample originated on flaws generated by ceramic inclusions from the casting mould at the subsurface of the part. Nevertheless, the 1B-HT sample remarkably resisted 9.85 million cycles at 615 MPa before experiencing failure. The fracture surfaces of alloys with 0.5 wt % B are shown in Figure 7. The 0.5B-AC alloy exhibited brittle striations that originated from the martensitic transformation induced by cyclic deformation at high crack growth [12,17]. The 0.5-HT alloy had an intergranular appearance at crack onset and during propagation and a transgranular appearance at the end. Figure 6 shows the fracture surfaces of alloys with 0.25 wt % B. The 0.25B-AC alloy exhibited a transgranular crack at the beginning that propagated in a radial direction, with dimples in the propagation zone and an intergranular final fracture. The 0.25B-HT alloy exhibited zones of faceted 'stair-like' deformation (brittle) and zones with dimples and microvoids (ductile). The fracture surfaces of alloys with 0.5 wt % B are shown in Figure 7. The 0.5B-AC alloy exhibited brittle striations that originated from the martensitic transformation induced by cyclic deformation at high crack growth [12,17]. The 0.5-HT alloy had an intergranular appearance at crack onset and during propagation and a transgranular appearance at the end. The fracture surfaces in alloys with 1 wt % B are shown in Figure 8. The analysed sample of 1B-AC alloy exhibited a transgranular crack at the beginning and in the propagation zone and intergranular appearance at the end; the crack nucleated on a boron carbide. The 1B-HT alloy was the most brittle among the studied alloys; a stair-like fracture surface is observed with pyramids and cleavages. The pyramids arose from the intersection of three planes in a secondary cleavage crack [27]. The fracture in this 1B-HT sample originated on flaws generated by ceramic inclusions from the casting mould at the subsurface of the part. Nevertheless, the 1B-HT sample remarkably resisted 9.85 million cycles at 615 MPa before experiencing failure. The fracture surfaces in alloys with 1 wt % B are shown in Figure 8. The analysed sample of 1B-AC alloy exhibited a transgranular crack at the beginning and in the propagation zone and intergranular appearance at the end; the crack nucleated on a boron carbide. The 1B-HT alloy was the most brittle among the studied alloys; a stair-like fracture surface is observed with pyramids and cleavages. The pyramids arose from the intersection of three planes in a secondary cleavage crack [27]. The fracture in this 1B-HT sample originated on flaws generated by ceramic inclusions from the casting mould at the subsurface of the part. Nevertheless, the 1B-HT sample remarkably resisted 9.85 million cycles at 615 MPa before experiencing failure.

Conclusions
Modifying the chemical composition of Co-based alloys via the addition of B increased the fatigue resistance in the as-cast condition according to the amount of B added; infinite fatigue life of 615 MPa was observed for the alloy with 1 wt % B; this represents an improvement of 60% regarding the base alloy. In contrast, heat treatment diminished the fatigue resistance of most alloys containing B, except the 0.06B-HT alloy, which retained its fatigue resistance at the same level of 433 MPa as observed in the as-cast condition. On analysing the fracture surfaces, the 0.06B-HT alloy exhibited greater ductility compared with the brittle behaviour observed in the other alloys even when ductile features were observed in the same samples. The 0.06B-HT alloy also exhibited a remarkable toughness to crack propagation; it endured considerably more cycles compared with any other alloy from the onset of macroscopic cracks until the final fracture.