Using Cavitation Peening to Improve the Fatigue Life of Titanium Alloy Ti-6Al-4V Manufactured by Electron Beam Melting

Although Electron Beam Melting (EBM) is an innovative technology, the fatigue properties of materials manufactured by EBM may be lower than those of casted and wrought materials due to defects and surface roughness. In order to enhance the fatigue life of components or structures manufactured by EBM, a mechanical surface treatment technology, e.g., peening, would be effective because peening introduces high compressive residual stress at the surface which can extend the fatigue life considerably. In the present study, specimens were manufactured by EBM using titanium alloy Ti-6Al-4V powder. Two types of specimens were prepared: as-built and as-machined specimens. Specimens of each type were treated by cavitation peening or shot peening. The fatigue lives of the specimens were evaluated by a plate bending fatigue tester. The residual stress and surface roughness were also evaluated. The results obtained showed that the fatigue strength of as-built specimens can be improved by 21% by cavitation peening or shot peening, and the fatigue life under particular applied stresses can also be extended by 178% by cavitation peening.

facturing of three dimensional components with complex geometries. Complete designs can be directly and freely done [1]. Due to these advantages, the demand for EBM in the aviation and medical industrial fields has risen. One of the advantages of EBM, which is due to its higher output and higher scanning velocity, is the rapidity with which three dimensional components can be manufactured compared to other additive manufacturing technologies, e.g., laser beam melting [2]. In contrast, it has been found that the fatigue performance of materials produced by additive manufacturing is lower than that of casted or wrought materials because of their porosity and surface roughness [3] [4]. For this reason, the fatigue properties of EBM materials are also no better than those manufactured by other additive manufacturing technologies [5]. Therefore, the fatigue properties of components manufactured by EBM are needed to be improved after manufacture.
To enhance the fatigue properties, some mechanical surface treatment technologies have been developed, e.g., shot peening and cavitation peening. Peening introduces compressive residual stress due to impact at the surface, inducing local plastic deformation. The residual stress in the surface layer has a significant effect on the fatigue properties, i.e., crack initiation and propagation, in metallic materials [6]. A tensile residual stress promotes fatigue fracture while a compressive residual stress does the opposite. In past studies, it has been demonstrated that the compressive residual stress introduced by peening prevents fatigue crack propagation in metallic materials [7]. The residual stress also affects chemical reactions on the surface. The resistance to corrosion can also be improved by controlling the residual stress [8].
Shot peening is a conventional mechanical surface treatment that utilizes the impact produced when solid bodies collide with the surface [9]. An innovative mechanical surface treatment, cavitation peening, utilizes the impact produced by shock waves in water caused by cavitation bubbles collapsing [10]. It can be utilized in the same way as shot peening. The cavitation bubbles are produced by injecting a high-speed water jet into water. Cavitation clouds spread out across the surface under treatment. When the bubbles collapse, they produce impacts which cause local plastic deformation on the surface. The intensity of the cavitation impacts can be controlled by several hydraulic parameters and the nozzle outlet geometry [11].
Cavitation peening has some advantages compared to conventional shot peening. Firstly, the surface roughness produced by cavitation peening is less than that produced by shot peening because there are no solid collisions [12]. Secondly, cavitation peening does not require the chemical cleaning which is needed to remove iron adhering to the surface when shot peening is applied, since cavitation peening uses only water. Thirdly, cavitation peening can be used to treat surfaces in narrow spaces where shot cannot reach, e.g., fastener holes in aircraft components [13], the bottoms of gears [14] or elements of the steel belts used for Continuously Variable Transmission (CVT) [15].
As mentioned above, the components manufactured by EBM need to be improved in terms of their fatigue properties, and cavitation peening has the potential to do this. In the present study, titanium alloy Ti-6Al-4V was employed, which was often used for high value-added products, e.g., medical products and/or aerospace products. In order to demonstrate this, plate bending fatigue tests were conducted on Ti-6Al-4V plate specimens manufactured by EBM after cavitation peening or shot peening. Additionally, as-built specimens and specimens whose surfaces were ground to remove surface roughness, i.e., as-machined specimens, were tested. The compressive residual stress at the surface introduced by peening was evaluated by X-ray diffraction stress measurements. Figure 1 shows a schematic diagram of the cavitation peening system used in the experiment [16]. Water stored in tank B is pressurized by a plunger pump with a maximum pressure of 35 MPa and maximum discharge of 3.0 × 10 −2 m 3 /min. The pressurized water is injected into a water filled chamber, i.e., tank A, through a nozzle immersed in water in the test section. A floating plate is located near the nozzle to avoid suction vortices. The nozzle is equipped with a cavitator and a guide pipe in order to enhance the aggressive intensity of the cavitating jet as shown in Figure 2. The geometry of the nozzle has been previously optimized [16]. The test section is open to the ambient air. As residual bubbles which remain after the cavitation bubbles have collapsed might reduce the shock wave, a partition plate is placed in tank B to remove residual bubbles remaining in the test water [17]. The geometrical parameters of the nozzle are given in Table 1. The injection pressure p 1 was kept constant at 30 MPa and the ambient pressure of the jet, p 2 , is equal to atmospheric pressure (0.1 MPa). The standoff distance from the upstream end of the nozzle plate to the surface of the specimen was 262 mm by reference to the   previous study [16]. The nozzle was scanned across the specimen at a constant velocity for uniform treatment. The processing time per unit length, t p , is defined by the scanning speed, v, and the number of scans, n, as follows:

Experimental Facilities and Procedures
In the present experiment, t p , was set to 10 s/mm for cavitation peening. The specimens were also treated by shot peening using recirculating shot accelerated by a water jet [18], as shown in Figure 3. The shot used in this study were JIS SUS440C metallic particles. The diameter and the number of shot were 3.2 mm and 500, respectively, and they were accelerated by a water jet with an injection pressure of 12 MPa through three orifices of 0.58 mm in diameter. The standoff distance from the orifice plate to the specimen was 50 mm. The head part shown in Figure 3(b) was scanned across the specimen along the guide by a motor. The processing time per unit length, t p , was set to 0.2 s/mm. In order to treat the specimens with the same aggressive intensity as cavitation peening, the processing time for shot peening was chosen as t p = 0.2 s/mm following evaluation of the arc height of an almen strip, which was 0.25 mm by cavitation peening with a processing time of t p = 10 s/mm.
All the specimens undergoing tests were manufactured by EBM. The powder for EBM was made of Ti-6Al-4V alloy and the size of the powder was about 75 μm in diameter. The spot size of the electron beam for selectively melting was 0.2 mm and the stacking pitch was 50 μm in the direction of the thickness. The as-built specimens were 90 mm long, 30 mm wide and 4mm thick as shown Figure 4. The width of center part was 20 mm with curvature radius of 45 mm. The surfaces of the as-built specimens were extremely rough, which may lead to cracks under fatigue cycling. After manufacturing the specimens by EBM, the surfaces of some specimens were ground to remove the roughness so that their fatigue lives could be compared with the as-built specimens. The ground specimens were denoted as "as-machined". The thickness of the "as-machined" specimens were 2 mm. Six types of specimen were prepared; As-built non-peened (NP), As-built cavitation peened (CP), As-built shot peened (SP), As-machined non-peened (NP), As-machined cavitation peened (CP), and As-machined shot peened (SP). In addition, the edges of all the specimens were polished by emery paper to prevent crack initiation from the edge.
A plate bending fatigue test is a suitable test for examining the effects of the surface mechanical properties on fatigue fracture. It has been demonstrated using plate bending tests that peening enhances the fatigue properties [19]. Therefore, a displacement controlled plate bending fatigue tester was used to evaluate the fatigue life of the specimens. All specimens were subject to bending stress with a stress ratio of R = −1. The frequency of the fatigue was set to 5 Hz for the first 10 6 cycles and then changed to 10 Hz beyond that. The residual stress at the surface was evaluated by a two-dimensional XRD (2D-XRD) method. The measurements were carried out using Cu Kα X-rays from a tube operated at 35 kV and 40 mA through a 0.8 mm diameter collimator. Diffractive Xrays from the Ti (2 1 3) lattice plane were detected using a two-dimensional position sensitive proportional counter with a diffractive angle of 140 degrees. Seventeen diffraction rings, i.e., Debye rings, were collected from several angles. The angle was chosen with reference to the previous study [20]. The exposure time was 120 s. Surface roughness measurements were conducted using a stylus with a tip diameter of 4 μm, a cutoff length of 0.8 mm, and a sampling length of 4 mm. The average surface roughness and maximum roughness were obtained from the average of five sets of experimental data. Figure 5 shows the amplitude of the bending stress, σ a , as a function of the number of cycles to failure, N f , for specimens in all conditions, i.e., the S-N curve. For the as-machined specimens, we focused on fatigue lives of less than 10 6 cycles. As shown in Figure 5, the fatigue life of the as-built specimen is lowest and both cavitation peening and shot peening extend its life considerably. The fatigue strength can be defined as the stress at which the fatigue life becomes beyond 10 7 cycles and it can be calculated by Little's method [21]. The fatigue strengths for the as-built NP, SP and CP specimens are 162. 1, 195.8 and 197.0 MPa, respectively. The fatigue strength has increased by almost 21% by cavitation peening and shot peening. In the case of the as-machined specimens, the straight lines are given by the following equation with the assumption that the gradients of the SP and CP specimens are the same as that of the NP specimen:

Results and Discussion
The gradient is obtained from the dataset of the as-machined NP specimen. C denotes the y-intercept, which is 1652 for NP, 1715 for SP and 1762 for CP. The fatigue life of the as-machined specimen has been extended by As shown in Figure 6, the estimated fatigue lives, N E , at σ a = 400 MPa are 106,444, 191,179, and 295,423 for the NP, SP, for CP specimens, respectively. The fatigue life has been extended by 80% and 178% by shot peening and cavitation peening, respectively. This effect can probably be attributed to compressive residual stress at the surface introduced by peening. Figure 7 shows the residual stress, σ R , in the bending direction at the surface of the as-built NP, SP, and CP and as-machined NP, SP, and CP specimens. The positive and negative values represent tension and compression, respectively. There is a tensile residual stress of 67 ± 18 MPa at the surface of the as-built NP specimen and this is mitigated and changed from tension to compression by peening. Compressive residual stresses of 127 ± 30 and 194 ± 34 MPa are introduced by shot peening and cavitation peening, respectively. There is a compressive residual stress of 220 ± 27 MPa in the as-machined NP specimen due to the machining. The value increases to 348 ± 26 MPa by shot peening and 648 ± 20 MPa by cavitation peening. Cavitation peening introduces high compressive residual stress since cavitation bubbles collapsing produce very high impact of the order of gigapascals and beyond, which are more severe than the impact produced by collisions between shot and the surface. Also, the strain velocity when the impact due to bubble collapsing deforms the surface is much higher than that produced by the impact of shot [22]. This difference gives rise to the differences in compressive residual stress and, as a result, the longer fatigue life. The diffractive X-ray profiles made for the stress measurements include much information about the microstructure. This is true, in particular, of the full width at half maximum (FWHM) of the profile, from which information about the micro-strain in the grains  can be obtained. The micro-strain represents the strain caused by adjacent grains and random strain in the grains. The FWHM in the as-built specimens are 1.128 deg for the NP specimen, 2.307 deg for the SP specimen, and 1.960 deg for the CP specimen. The value increases with peening since the melted surface is work hardened by the impact produced. That is, many dislocations are introduced by work hardening and these are refined as a result of the peening. However, the values in the as-machined specimens are 1.692 deg for the NP specimen, 1.861 deg for the SP specimen, and 1.662 deg for the CP specimen, which, in comparison with the as-built NP specimen have increased, which is a consequence of the work hardening induced by the machining. Interestingly, cavitation peening decreases the FWHM as opposed to shot peening even though a compressive residual stress is also introduced. In past literature, it has been shown that cavitation peening can decrease the FWHM while also introducing compressive residual stresses in titanium alloy rods made of Ti-6Al-4V [23]. It is likely that the ultrasound produced by cavitation bubbles collapsing moves dislocations, which may be the source of the microstrain [24]. Since it is possible that the micro-strain is the origin for crack initiation, the decrease in the FWHM can be seen as a beneficial effect for improving the fatigue life.
In general, one of the most important factors affecting the fatigue life in addition to the residual stress is surface roughness [25]. A roughened surface promotes crack initiation due to stress concentration. Figure 8 shows the surfaces of each of the specimens. Moreover, the arithmetic average roughness, Ra, and maximum roughness, Rz, are shown in Figure 9. Particles stacked by EBM can be seen at the surface of the as-built NP specimen as shown in Figure 8(a). These particles make the surface extremely rough. Ra and Rz are 16.29 ± 1.85 and 83.53 ± 5.23 μm, respectively, for the as-built NP specimen. Stacked particles can also be observed in the CP specimen whereasthe impact of the shot in shot peening makes the surface flatter. In cavitation peening, cavitation bubbles enter the gaps between the particles and introduce compressive residual stress at the bottoms of the gaps. This isshown by the results of the roughness measurements; (Ra, Rz) = (7.15 ± 0.89, 41.19 ± 6.51) μm for the SP specimen and (16.13 ± 1.82, 82.24 ± 2.83) μm for the CP specimen. Since, unlike cavitation bubbles, shot cannot reach the bottoms of the gaps between the particles where the fatigue cracks initiate, the difference in roughness between the SP and CP specimens does not correlate with the increase in fatigue life. This supports the fact that the introduction of compressive residual stress into the as-built specimen by shot peening is less than that by cavitation peening. In the as-machined specimens, the surface roughness is dramatically improved by the machining; (Ra, Rz) = (0.22 ± 0.03, 1.51 ± 0.40) μm for the NP specimen. The surface roughness of the CP specimen is higher than that of the SP specimen; (Ra, Rz) = (0.32 ± 0.02, 2.23 ± 0.461) μm for the SP specimen and (0.53 ± 0.08, 3.14 ± 0.38) μm for the CP specimen. This is because the impact due to cavitation bubbles collapsing forms local, deep plastic deformation pits at the surface even though the surface has a high yield stress. In contrast, wide and shallow deformation marks can be observed in the SP specimen. That is, the impact of shot in this experiment was unable to induce large plastic deformation in the sub-surface of the Ti-6Al-4V specimens. It is quite likely that the difference in the compressive residual stress can be attributed to the differences between the plastic deformation pits. In order to examine the combined effect of the compressive residual stress and surface roughness on fatigue failure, we introduce a parameter, * R σ , given by Equation (4) [26], in which the compressive residual stress has a positive effect and the surface roughness has a negative effect on the fatigue life: where σ CR denotes the compressive residual stress at the surface. Other than the value of the compressive residual stress, another factor that has a positive effecton fatigue life is the depth of the compressive residual stress. Thus, the square of the compressive residual stress is included in * R σ . Figure10 plots the estimated fatigue life at σ a = 400 MPa, N E , versus the parameter * R σ for as-machined specimens. As shown in Figure 10, the fatigue life increases linearly with * R σ . Thus, this parameter is a good indicator for the fatigue life. That is, the beneficial effects of peening on the fatigue life can be indicated by considering the surface roughness and compressive residual stress.
Finally, Figure 11 shows cross sections of the fractured surfaces of the specimens. It seems that small cracks initiate at various points near the surface, propagate along the surface and into the depth, e.g., semi-elliptical in shape, and coalesce with each other. After they coalescence, a high stress intensity factor, which represents the magnitude of the stress field, around a large crack leads to unstable and rapid fatigue crack propagation. The compressive residual stress mitigates the actual applied stress and also causes a crack closure effect [27] [28] when cracks occur at the surface, since the residual stress affects the stress field around the crack tip, i.e., the stress intensity factor [29]. Therefore the stress field near the surface and around the crack tip after crack initiation can be greatly reduced by compressive residual stress. Although components and structures manufactured by EBM have low resistance to fatigue failure, peening, especially cavitation peening, can greatly increase it and

Conclusions
In order to demonstrate that the fatigue life of metallic material manufactured by Electron beam melting (EBM) can be improved by cavitation peening, specimens manufactured by EBM using Ti-6Al-4V powder were treated by cavitation peening. For comparison, specimens were also treated with shot peening. Two types of specimens, as-built specimens and as-machined specimens, were prepared. The as-machined specimens were prepared by removing the surface roughness, which was due to stacked particles, from the as-built specimens by machining. The fatigue life, i.e., the S-N curve, was evaluated using a displacement controlled plate bending fatigue tester, and the residual stress at the surface was measured by an X-ray diffraction method. Moreover, the surface roughness, i.e., the average roughness, Ra, and the maximum roughness, Rz, was measured by a scanning stylus method. The results obtained are summarized as follows: 1) Specimens manufactured by EBM have low fatigue resistance and their fatigue lives can be improved by cavitation peening more effectively than by shot peening. In the present experiments, the fatigue strength, which is the applied stress at which the fatigue life goes beyond 10 7 cycles for the as-built specimens, is increased from 162 MPa to 195 MPa by shot peening and to 197 MPa by cavitation peening. For the as-machined specimens, the fatigue life at an applied stress of 400 MPa was increased by 80% by shot peening and 178% by cavitation peening. 2) Cavitation peening introduced a compressive residual stress of 194 ± 34 MPa even though the surface of the as-built specimen was extremely rough (Ra = 16.29 ± 1.85 μm, Rz = 83.53 ± 5.23 μm). Shot peening introduced a compressive residual stress of 127 ± 30 MPa in the as-built specimen. Regarding the as-machined specimens (Ra = 0.22 ± 0.03 μm, Rz = 1.51 ± 0.40 μm), shot peening and cavitation peening introduced compressive residual stresses of 348 ± 26 and 648 ± 20 MPa, respectively. With regard to cavitation peening the full width at half maximum (FWHM) in the X-ray diffraction profile, which was related to the microstrain, decreased from 1.692 deg to 1.662 deg, while further work hardening was introduced. In contrast shot peening increased the FWHM to 1.861 deg. It is possible that the increase in micro-strain is the origin of the crack initiation.
3) The effect of peening on fatigue can be represented by a parameter that depends on a combination of the compressive residual stress, which has a positive effect, and the surface roughness, which has a negative effect. In the present study, the increase in fatigue life was found to have a linear relationship with the parameter.