Mechanical and corrosion fatigue behaviors of gradient structured 7B50-T7751 aluminum alloy processed via ultrasonic surface rolling

1. Introduction

Owing to light weight and high specific strength, 7B50 aluminum alloy is used in the manufacture of load-bearing structural components for the aviation, aerospace, and automotive industries [1,2]. Most failures, such as fatigue fracture, wear, and corrosion, originate from the outer layers of engineering structural components. These failures, which are closely correlated with the surface microstructure and properties of materials, can be effectively restrained by one or more surface treatment techniques [3]. Shot peening is a common surface treatment which raises the fatigue or corrosion fatigue resistance of engineering structural components and its strengthening mechanism has been widely studied [4,5]. Recently, surface severe plastic deformation (SSPD) techniques (including surface mechanical attrition treatment [6,7], ultrasonic shot peening (USSP) [8], ultrasonic impact treatment (UIT) [9], deep rolling (DR) [10,11], surface mechanical grinding treatment (SMGT) [12], and surface mechanical rolling treatment (SMRT) [13]) have received considerable attention due to their simplicity and ability to produce gradient structured (GS) metallic materials.

GS materials are characterized by a gradient distribution of grain sizes (from nanoscale to microscale) [14]. Based on the grain size and defect distributions, these materials generally contain the following layers [15]: an outermost nanolaminated or nanograined (NG) layer (mean grain size: <100 nm); an ultrafine laminated or ultrafine grained (UFG) layer (mean grain size: 100-500 nm); a deformed layer with submicron- and micron-sized grains; and internal undeformed coarse grains. The grain refinement mechanism of metallic materials is closely correlated with their stacking fault energy (SFE). Grain size evolution in aluminum and its alloys, which have a high SFE (~0.2 J/m²), is controlled by dislocation activities [6]. Previous studies [[16], [17], [18]] have reported that GS materials, which possess a hard NG or UFG outer layer and soft internal layers, exhibit unique mechanical behaviors, such as extraordinary tensile ductility and significant strain hardening. Fatigue crack initiation would be suppressed by the NG or UFG layer in the gradient structure, and crack propagation would be hindered by the internal coarse grains [19]. Furthermore, the surface chemical activity of UFG materials is high, owing to the small grain size, and this activity facilitates enhanced re-passivation when metastable pits are formed [20]. Krishna et al. [21] investigated the localized corrosion behavior of UFG Al-4Zn-2Mg alloy and reported that its high corrosion resistance results mainly from ultrafine grains, fine deleterious particles, and the absence of a precipitate-free zone.

The ultrasonic surface rolling process (USRP), which combines the advantages of UIT, DR, SMGT, and SMRT, is a recently developed SSPD technique [22,23]. The ultrasonic rolling component is installed on a computer numerically controlled lathe, allowing the sample surface to be homogeneously treated through accurate control of the process parameters [24]. Some materials subjected to a USRP treatment have been investigated. For example, Liu et al. [25] reported the nanocrystallization deformation behavior and mechanism of a body-centered tetragonal-structured 17-4 precipitation hardening stainless steel after 30-pass USRP. Li et al. [26] stated that a multi-pass USRP treatment resulted in improved mechanical properties and fatigue resistance of a Ti-6Al-4 V alloy. Our previous work [27] assessed the effect of USRP on the corrosion fatigue resistance of 7B50-T7751 aluminum alloy without gradient structure, especially the samples subjected to single pass treatment. However, the corrosion fatigue behavior of this alloy with gradient structure remains unclear, and systematic investigations of the mechanical properties characterizing this alloy are lacking. Therefore, the purpose of the present work is to prepare a GS layer on the 7B50-T7751 aluminum alloy via 12-pass USRP. In addition, we evaluate the mechanical properties, including the tensile strength, ductility, toughness, and strain hardening behavior, as well as the corrosion fatigue behavior of the GS aluminum alloy. We clarify the influence of the surface roughness, grain refinement, and compressive residual stress on the aforementioned properties. Furthermore, we provide guidelines for further development of USRP that yields improved practical characteristics of 7000 series aluminum alloys.

2. Experimental

2.1. Material and sample processing

A commercially rolled 7B50-T7751 aluminum alloy plate (thickness: 40 mm) was used as the raw material in the present work, with the chemical composition (wt%) of 6.2 Zn, 2.1 Mg, 2.0 Cu, 0.08 Zr, 0.05 Ti, 0.04 Si, 0.03 Fe, and balance Al. In accordance with ASTM standard E8 [28], round specimens for tensile testing were cut along the rolling direction (gauge length: 50 mm, diameter: 10 mm). Similarly, in accordance with ASTM standard E466 [29], specimens with circular cross-sections for corrosion fatigue testing were prepared along the rolling direction. The diameter and length of the test section were 10 mm and 35 mm, respectively.

A schematic of the USRP device has been provided in Ref. [27]. During the USRP, the aforementioned 7B50 samples rotated at a velocity of 75 rev/min. An ultrasonic shock vibration (amplitude: 10 μm; frequency: 28 kHz) and static force (500 N) were simultaneously applied along the normal direction of the sample surface via a freely rotating tungsten carbide/cobalt (WC/Co) ball. The ball slid along the axial direction of the sample at a feeding rate of 0.14 mm/rev. The samples experienced one pass when the ball slid from one end of the surface to the other end. A gradient structure was generated by subjecting the samples to 12 passes (the corresponding samples were referred to as UR12). The untreated samples were referred to as BM. Furthermore, cylindrical samples (length: 15 mm) for microstructural characterization and electrochemical testing were cut from the test section of the fatigue samples before and after USRP.

2.2. Microstructural characterization

A DH300LCS optical microscope (OM) was employed to observe the cross section of UR12 sample etched by Keller’s reagent. The detailed cross section and surface morphology of this sample were investigated via scanning electron microscopy (SEM: Tescan Vega II). The corresponding mean surface roughness (S_a) was calculated from the data collected by means of confocal laser scanning microscopy (CLSM: LSM 700) and the calculation formula can be found in Ref. [30]. Furthermore, the microstructure comprising the surface layer of UR12 sample was characterized via transmission electron microscopy (TEM: Tecnai G² F20) with an operating voltage of 200 kV. Subsurface layers at different depths were obtained via gradual mechanical grinding from the topmost surface. Afterward, these layers were subjected to single-sided ion-beam milling for TEM observation. Crystal plane arrangements of UR12 sample were analyzed by using X-ray diffraction (XRD) spectroscopy (D/max 2500 apparatus with CuK_α radiation).

Residual stress of UR12 sample was measured, via the sin²ψ method, using an X-ray stress analyzer (MSF-3 M apparatus with CrK_α radiation). The normal angles of ψ were 0°, 18.40°, 26.60°, 33.20°, 39.20°, and 45.00°. The {311} diffraction peak and 139.3° diffraction angle of aluminum alloy were chosen for this measurement. To determine the longitudinal residual stress at different depth of radius, the surface layer of a cylindrical sample was sequentially removed by means of chemical stripping, as described in our previous study [27]. The measured values of the residual stress were rectified in accordance with the SAE standard [31].

2.3. Mechanical testing

The HV-1000 microhardness tester was utilized to measure the in-depth microhardness profile of UR12 sample, equipping with a Knoop indenter. The indentation load and the holding time were 0.245 N and 20 s respectively. The aforementioned chemical stripping method was also employed to remove surface material step by step during this measurement.

Tensile testing was conducted on the Instron 3382 Floor Model Universal Testing System with samples loaded at a tensile speed of 1 mm/min while the corresponding engineering stress-strain (σ-ε) curve of each sample was obtained. The part before necking of the σ-ε curve was converted into a true stress-strain (σ_T-ε_T) curve by the uniform deformation model. Assuming constant volume prior to necking, the relation between ε, ε_T, σ, and σ_T can be expressed as follows [32]:

ε_T = ln(1+ε) (1)

σ_T = σ(1+ε) (2)

2.4. Corrosion fatigue testing

Stress-controlled axial corrosion fatigue (CF) testing was performed on a SDS110 servo hydraulic system, and the stress referred to the stress amplitude (σ_a), i.e., the (maximum stress-minimum stress)/2. The test portion of each CF sample was immersed in a Plexiglas container filled with the 3.5% NaCl neutral aqueous solution at 25 °C [33]. The sinusoidal cyclic frequency and the stress ratio (R) were 10 Hz and 0.1 respectively in the testing. The testing was terminated when the specimens fractured or 10⁶ cycles were exceeded. Moreover, some specimens (denoted as UR12-R) were subjected to a pre-fatigue process prior to the CF testing. This process was the fatigue in air under a σ_a of 157.5 MPa (R = -1), which ran 10 cycles with a 0.1 Hz sinusoidal wave. In addition, the fracture surface and corroded surface of typical CF samples were examined via SEM.

2.5. Electrochemical testing

Using a PARSTAT 2273 electrochemical station connected to a three-electrode cell, the electrochemical tests were performed in a neutral 3.5% NaCl solution at 25 °C. The cylindrical sample (exposed area: 0.5 cm²), saturated calomel electrode (SCE), and thin platinum foil were used as the working, reference, and counter electrodes, respectively. When the open circuit potential (OCP) was stable, polarization curves of different samples were acquired at a scanning rate of 0.5 mV/s, and potential ranging from -0.25 V_SCE to 0.25 V_SCE with respect to the free corrosion potential (E_corr).

3. Results and discussion

3.1. Microstructure of the GS layer

The cross section of the UR12 sample is displayed in Fig. 1(a). Due to the synergism of shear stress and ultrasonic shock vibration provided by USRP, the T-oriented grains near surface deflected along the shear direction and were elongated, yielding a plastic flow feature. A gradient deformed layer (thickness: ~425 μm), including a severe plastic deformation (SPD) layer (thickness: ~190 μm) was formed on the UR12 sample, and a micro-crack (depth: ~32 μm) was also observed (see inset). The UR12 sample exhibited a scaly surface morphology with a mean surface roughness of S_a =1.30 μm, which was considerably larger than that (0.691 μm) of the untreated sample [27]. Compared with other regions, the inhomogeneous regions of the surface (e.g., micro-cracks, pores, passive-film defects) are more easily penetrated by water molecules, oxygen, and chlorine ions and, hence, may be more susceptible to localized corrosion attack. However, this does not always hold true for the modified surface [30]. Corrosion resistance of the UR12 sample will be analyzed later. A large amount of shear strain was accumulated at the near surface layer of the sample after each pass of USRP. When the strain reached a critical value, the ductility exhaustion resulted in the surface failure of the treated sample. This failure was manifested as (i) material separation from the surface and, hence, debris generation (wear), and (ii) micro-crack initiation and subsequent propagation (rolling contact fatigue) [34,35]. The surface morphology, surface roughness increment, and micro-crack initiation of the alloy after 12-pass USRP are attributed to the aforementioned factors.

The bright-field (BF) and corresponding dark-field (DF) TEM images shown in Fig. 2 reveal details of the microstructure occurring at different depths in the UR12 sample. Fig. 2(a) exhibits that many spherical G.P. zones and some strip-like matrix precipitates (η'-MgZn₂) [36] were homogeneously distributed in the grains of core layer (5 mm deep from the top surface). Besides, discontinuously distributed grain boundary precipitates (η-MgZn₂) [37] can also be clearly observed. The above precipitates were all on the nanometer scale. Fig. 2(b) shows that dislocation walls (DWs) and dislocation tangles (DTs) were formed in the grains located at a depth of 350 μm from the top surface. The formation of DWs and DTs resulted from dislocation multiplication caused by the activation of various dislocation sources (e.g., Frank-Read sources) [38]. As indicated by the black arrow, the DWs led to a sub-division of the original grain. Moreover, the inset reveals fine dispersed G.P. zones and η'-MgZn₂ as well as η-MgZn₂. The degree of the DWs-induced sub-division was especially severe in the SPD layer (150 μm deep from the top surface), as shown in Fig. 2(c). The DF TEM image also revealed a small number of matrix precipitates and grain boundary precipitates. At a depth of 40 μm from the top surface, coarse grains have been sub-divided into fine grains (see Fig. 2(d)). Simultaneously, few precipitates were visible in the DF image. The ring-like selected area electron diffraction (SAED) pattern (see Fig. 2(e)) corresponding to Fig. 2(d) and the calibrated Miller indices confirmed that the area contains many aluminum grains with high angle boundaries. Fig. 2(f) shows the grain size distribution which is 40 μm deep from the top surface, revealing that the average grain size is ~67 nm. These TEM observations indicated that a gradient structure was successfully introduced into the UR12 sample. Microstructure refinement was a typical feature of the SPD layer, as reported in other authors [8,25,39]. For example, Wu et al. [8] investigated the microstructures comprising the surface layer of a USSP-treated 7075 aluminum alloy. The results revealed that the ultrafine grained structures were generated by introducing intense strains and high strain rates into the 62 μm thick surface layer. The plastic strain was accommodated via grain sub-division into subgrains, i.e., the main mechanism for grain refinement in that study.

The XRD patterns of the BM and UR12 samples are exhibited in Fig. 3. It can be seen that the preferential orientation of the UR12 sample had a tendency to become (111) from (220). This resulted from the fact that at least two sets of slip systems of the aluminum were actuated to produce the requested strain when the shear deformation occurred, and {111} <112>, {111} <110>, {112} <110>, and {001} <110> were the main slip systems [40]. Fig. 3 also reveals peak broadening and shifting, especially for (222) Bragg diffraction peak. Considering the principle of XRD and the uniform intrinsic broadening effect of a given tester, the peak broadening of the UR12 sample resulted mainly from microstrain and grain refinement. Similar results have been reported for other alloys subjected to SSPD [41,42]. Furthermore, Sun et al. [41] indicated that the interplanar distance of 7150 alloy increased, owing mainly to the USSP-induced η phase dissolution. This resulted in peak left-shifting which referred to the positions of Bragg diffraction peaks shifted to lower 2θ values, thereby significantly suppressing the peak right-shifting caused by the introduced compressive residual stress. Therefore, the XRD peak left-shifting of the USRP-treated 7B50 alloy may have resulted from an increase in the interplanar distance of Al due to the dissolution of nanoscale precipitates (see Fig. 2). It is worth noting that the microscale Fe-rich intermetallic precipitates identified by our previous study [37] still can be observed in the nanograined layer (see Fig. 1(a)). This indicates that USRP did not lead to the dissolution of all types of precipitates.

3.2. Surface corrosion performance of the GS layer

The microstructure analysis of the GS layer showed that micro-cracks occurred in the surface layer of UR12 sample and, hence (for the sake of comparison), the UR12-P sample (UR12 sample after removal of 40 μm thick surface layer) was taken as the crack-free control group. The chemical stripping method (see Section 2.2) and silicon carbide sandpaper (#3000) polishing were used for sequential removal of this layer. The UR12-P sample subjected to a pre-fatigue process denoted as UR12-PR sample. The corrosion testing solution was stabilized in 20 min, and then the OCP values of BM, UR12, UR12-R, UR12-P and UR12-PR samples were determined via 10 min measurements. Fig. 4(a) shows that the OCP values of each sample stabilized after 8 min of testing, and the values of USRP-treated samples were more positive than that of BM sample. This was especially true for the UR12-P sample that was characterized by low surface roughness (S_a: ~0.2 μm). The USRP-induced OCP shifting resulted probably from variations in the surface passive films and surface micro-roughening in the treated area, consistent with the findings of previous studies [30,43,44]. For example, Lv et al. [43] reported that surface passive films on burnished 2024 aluminum alloy samples were considerably denser and more protective than those on burnish-free samples, and led to positive OCP shifting. Peyre et al. [44] proposed that the effective passive layer was difficult to form on a 316 L stainless steel due to the inhomogeneous surface micro-roughening after laser peening.

Fig. 4(b) shows the polarization curves measured for the different samples. The cathodic branch and the anodic branch of the curves corresponded to the oxygen reduction reaction and the active dissolution reaction, respectively, of the 7B50 aluminum alloy. According to Ref. [45], the corrosion current density (i_corr) of each sample can be determined via extrapolation of the Tafel slope. Furthermore, the corrosion rate (CR; μm/year) of the 7B50 aluminum alloy was calculated in accordance with Faraday’s law [46] and Ref. [30]:

CR = 10.89·i_corr (3)

The values of E_corr, i_corr, and CR were calculated for the BM, UR12, UR12-R, UR12-P, and UR12-PR samples (see Table 1). As the table shows, the E_corr values of USRP-treated samples were more positive than that of BM sample, similar to the OCP shifting after USRP. Other authors have reported similar results for the positive E_corr shifting, due to modification of the thin aluminum oxide film and introduction of a compressive residual stress, resulting from a surface treatment [30,47]. The order of i_corr can be described as: UR12-P<UR12<UR12-PR<UR12-R<BM, which was also consistent with the order of E_corr. The calculated CR values of UR12-P and UR12 samples were reduced by 69.77% and 60.08% compared with that of the untreated sample. Although UR12 sample had a rough surface, its corrosion resistance was only slightly lower than that of UR12-P. Besides, other USRPed samples also presented the low values of CR. The aforementioned results indicated that the inhibited anodic dissolution reactions in the USRPed samples exposed to the aggressive NaCl solution led to a significant decrease in the CR value.

3.3. Mechanical properties of the USRP sample

3.3.1. Microhardness

Fig. 5 shows the in-depth microhardness profile of the UR12 sample. After 12-pass USRP, a gradient work hardening layer (~225 μm deep from the topmost surface) was formed on the 7B50 alloy. The microhardness (mean value: ~3.9 GPa) decreased to the matrix value of 1.5 GPa with increasing depth, i.e., the surface microhardness of UR12 sample was increased by 160%. This trend was generally correlated with the gradient change in dislocation density and microstructure refinement. Sangid [48] indicated that persistent slip bands (PSBs) represent a critical mechanism of fatigue crack initiation in ductile face-centered cubic materials. Huang et al. [13] proposed that the SMRT-introduced work hardening hindered the formation of PSBs on the AISI 316 L stainless steel, thereby inhibiting fatigue crack initiation. This suggested that the USRP-induced increase in microhardness is favorable for enhancing the CF resistance of the UR12 sample.

3.3.2. Tensile properties

The σ-ε curves and calculated σ_T-ε_T curves (before necking) of the BM, UR12 and UR12-P samples are shown in Fig. 6(a) and (b). The mechanical properties (e.g., the initial yield stress (σ₀), 0.2% offset proof stress (σ_0.2), ultimate tensile strength (UTS), and tensile ductility (ε_f)) of the three samples can be directly obtained from the σ-ε curves. Furthermore, the toughness (U_T) was determined by integrating the σ-ε curves. The corresponding strain hardening exponent (n) was derived by fitting the uniform plastic deformation region of the σ_T-ε_T curves (ranging from point B to point C) using Ludwik’s equation [49]:

σ_T=K₁+K₂$ε_T^n$ (4)

where K₁ and K₂ are the initial yield true stress and the strengthening coefficient, respectively. According to Ref. [50], the strain hardening rate θ in the homogeneous plastic deformation region was estimated using the equation transformed from Eq. (4):

$θ=\frac{dσ_T}{dε_T}=n\frac{σ_T}{ε_T} $ (5)

The aforementioned mechanical properties for the studied samples are given in Table 2. Compared with those of the BM sample, the σ₀ and UTS of the UR12 sample were 8.27% and 2.01%, higher, respectively, whereas the tensile ductility was 6.53% lower. The higher strength of the UR12 sample was attributed to the induced dislocation strengthening and grain refinement in the gradient structured layer, consistent with the microstructural features resulting from other SSPD methods [13,51]. The aforementioned lower tensile ductility of this sample may be attributed to the fact that: strain hardening inhibited the motion of dislocations [52], and the weak surface (micro-cracks in the surface layer) acted as a suitable site for crack initiation and propagation during tensile testing. This was verified by the almost complete restoration of the tensile ductility characterizing the UR12-P sample. In addition, the σ_0.2 values of the UR12 and UR12-P samples reduced by 2.03% and 3.12%, respectively, compared with that of the BM sample. There was a slight decrease in σ_0.2 in the gradient structured samples during uniform plastic deformation. This may be related to the competition between the two factors namely, the strain hardening of GS 7B50 Al alloy and, the dissolution of nanoscale precipitates in the treated samples (see Fig. 2). The high strain induced by USRP increased the dislocation density and decreased the grain size, which was beneficial to raising the yield strength [53]. Meanwhile, USRP led to the massive dissolution of nanoscale precipitates in the nanograined layer. These precipitates provided the dislocation motion resistance are fundamental to the precipitation hardening mechanism in ultrahigh strength aluminum alloys [53]. Obviously, the dissolution of nanoscale precipitates was not conducive to improving the yield strength. The result of competition between the aforementioned two factors indicated that aluminum alloys subjected to USRP needed to consider the influence of precipitate changes on their mechanical properties. Other authors have reported similar results of σ_0.2 reduction in USRP-treated materials [26,54]. For example, Ao et al. [54] reported that the tensile strength increased slightly, but the yield strength (σ_0.2) decreased significantly in USRP-treated Ti-6Al-4 V alloy. This was ascribed to the induced gradient structured surface layer and the phase transformation, causing the easy dislocation slip during plastic deformation.

Generally, an enhancement in the strength of the material is available at the expense of the ductility [54]. The 7B50-T7751 aluminum alloy was characterized by low ductility and high strength. And a GS layer only constituted a small fraction (~8.5%) of the 10-mm-diameter sample. Therefore, the deformation strengthening of the alloy after 12-pass USRP was not particularly significant. However, the strengthening effect of a high ductility 304 stainless steel subjected to a similar surface treatment (ultrasonic nano-crystal surface modification) was substantial [51].

The strain hardening behavior of a material can be characterized via the strain hardening exponent. The n value of the UR12 sample was 26.32% higher than that of the untreated sample. This comparison indicated that 7B50 alloy with gradient structure exhibited a high resistance to continuous plastic deformation. Similarly, the n value of the UR12-P sample, with a thinner GS layer (thickness: ~385 μm) than the UR12 sample, was also greater than that of the BM sample. These trends are similar to those reported in Ref. [15]. The mechanical behaviors of different gradient structured layers in nickel alloy subjected to SMGT were investigated in that work [15], and the results revealed that these layers had high n values. That is, the gradient strain introduced geometrically necessary dislocations [55] (separate from the statistically stored dislocations) for maintaining the crystal continuity and accommodating the deformation incompatibility with the adjacent layer. This enabled significant strain hardening of the alloy. Fig. 6(c) shows the strain hardening rate as a function of the true strain associated with the uniform plastic deformation region of the investigated samples. Similar trends were found for the θ of the untreated and the treated samples, i.e., the strain hardening rate was high in the initial stage of deformation and then decreased continuously with increasing true strain. Moreover, the θ values of the UR12 and UR12-P samples were higher than that of the BM sample. The initial strain hardening rate of the USRP-treated samples increased due to the grain refinement. Gashti et al. [38] attributed the increase in θ to the relatively high number of dislocation sources, the dislocation interactions, and the short traveling distances of dislocations prior to meeting grain boundaries. The aforementioned trends observed for n and θ indicated that the 12-pass USRP had a notable effect on the strain hardening behavior of the 7B50-T7751 Al alloy.

Toughness represents a balance of the ductility and strength of a material. Table 2 shows that U_T values of UR12 and UR12-P samples were 4.99% and 2.40% lower than that of untreated sample respectively, indicating that the USRP treatment had a slight adverse effect on the toughness of the treated samples.

3.4. Corrosion fatigue properties of the USRP sample

3.4.1. CF strength and fatigue strength exponent

The CF behavior of the 7B50 aluminum alloy with or without USRP treatment can be seen from Fig. 7 which is a double logarithmic plot. The fatigue behavior of metallic materials is typically divided into two dissimilar stages: the stage encompassing long crack propagation in low-cycle fatigue (LCF) and the stage prior to short crack initiation in high-cycle fatigue (HCF) [56]. In our research, the number of cycles to failure (N_f) below and above 5 × 10⁴ were considered the LCF and the HCF regions, respectively. CF strength (stress amplitude (σ_a) corresponding to a CF life of 10⁶ cycles) values of 67.5 and 146.25 MPa were obtained for the BM and UR12 samples, respectively. This indicated that the USRP treatment increased the corrosion fatigue strength of 7B50 alloy by 117%. The relationship between σ_a and N_f in the HCF region is typically described by Basquin’s formula [13]:

σ_a=σ'_f (2N_f)^b(6)

where σ'_f and b refer to the fatigue strength coefficient and the fatigue strength exponent respectively.

This equation yielded a strong linear dependence of σ_a on N_f (see Fig. 7). Through regression analysis, σ'_f and b values of 1023 MPa and-0.1846 as well as 1231 MPa and-0.1472 were obtained for the BM samples and the UR12 samples, respectively. Xin et al. [39] and Li et al. [57] revealed that σ'_f may be correlated with the tensile strength. The results also revealed that b is influenced by strain localization and the stress gradient during crack initiation and crack propagation, respectively. According to the authors [57], a gradient nanostructured layer introduced by SMRT constitutes an ideal microstructure for suppressing strain localization and hindering fatigue crack propagation, thereby leading to an increase in b under stress-controlled fatigue. This may explain the change in b of the UR12 samples with a similar GS layer. However, the correlation between σ'_f and UTS in our study was weaker than that observed in some previous studies [13,39]. This is attributed to the fact that the interaction between the environment and the cyclic load resulted in extremely complex CF behavior of the metallic materials.

3.4.2. Influence of CRS and microstructure on CF

The USRP-induced axial residual stress distributions are shown in Fig. 8, where the negative and the positive values represent the compressive residual stress (CRS) and the tensile residual stress respectively. The CRS in the UR12 sample reached a value of approximately 133.2 MPa at the topmost surface and increased gradually to 191.5 MPa at a depth of 86 μm from the surface. Thereafter, the CRS decreased gradually with increasing depth and then transformed into a tensile residual stress. The CRS distribution extended to a depth of ~700 μm. The value and distribution of the CRS in the UR12 sample were both greater than those of the untreated sample. Furthermore, the USRP-induced CRS distribution of the 7B50 alloy was considerably greater than the distribution resulting from other surface treatments such as shot peening and ball-burnishing [4]. Rodopoulos’s research group [58,59] found that stress concentrations caused by surface roughening after controlled shot peening can facilitate the initiation and early propagation of fatigue cracks in high strength aluminum alloys, which was compensated by crack closure acquired from the induced CRS. Once the compensation surpassed a certain boundary condition, the fatigue resistance of studied aluminum alloys was vastly enhanced. This finding can be employed to explain significantly higher CF strength of the UR12 sample with a rough surface (compared with that of the BM sample).

Fig. 9 shows the comparison of CF life of different samples under the same stress, evaluating the synergistic effect of the CRS and microstructure on the corrosion fatigue resistance of these samples. Our previous study showed that the inherent CF behavior of 7B50 alloy was slightly affected by the pre-fatigue process [27]. The mean CF life of UR12-R sample (i.e., 219362 cycles) was 61.38% lower than that (567969 cycles) of the UR12 sample. For the UR12-R sample, the CRS at the surface, maximal CRS, distance between the location of maximal CRS and the surface, and total depth of the CRS distribution were 67.76 MPa, 137.5 MPa, 44 μm, and 625 μm, respectively (see Fig. 8). Compared with UR12 sample, the above four CRS parameters were decreased by 49.1%, 28.2%, 48.8% and 10.7% respectively. The CRS of UR12-R sample was significantly relaxed and this relaxation led to an apparent reduction in the extent of improvement in CF. After removing the partial surface layer, the average CF life of UR12-P samples with low surface roughness reached 793135 cycles. This CF life was ~27 times higher and 110% higher than those of the BM (28330 cycles) and the UR12-PR sample (377656 cycles), respectively. The comparison of mean corrosion fatigue life among different samples further explained the effect of CRS and roughness on the CF behavior of GS 7B50 alloy, indicating that CRS was the dominant beneficial factor for the improvement of CF resistance.

Furthermore, our previous research [27] showed that the surface CRS and the average CF life of one-pass USRPed 7B50 alloy without GS layer were deceased by 44.65% (from 219.7 to 121.6 MPa) and 83.24% (from 778025 to 130395 cycles) respectively after the relaxation of CRS. This indicated that the CF life of one-pass USRPed sample had a strong dependence on the introduced CRS. The value of the CRS associated with the UR12-P sample (see Fig. 8) in this work was smaller than that of the one-pass USRPed sample, and the average CF life and surface roughness of the two types of samples were almost the same. However, the mean CF life of the UR12-P samples decreased by only 52.38% (from 793135 to 377656 cycles) after the stress relaxation, which showed a reduced extent of decline in CF life. This difference in 7B50 alloy with and without GS layer indicated that the gradient structure was an important factor and contributed synergistically to the improvement of CF resistance. Because the gradient increase in grain size from the surface to the sub-surface is considered favorable for increasing the initiation threshold of fatigue crack and reducing the growth rate of fatigue crack [13,60].

3.4.3. Analysis of fractography and surface morphology

Fig. 10 shows the CF fractography of different samples. The CF crack initiation in the 7B50-T7751 aluminum alloy occurred via pitting (see Fig. 10(a)), as confirmed in our previous study [27]. Furthermore, the fractures of the UR12 and UR12-P samples originated at a depth of ~1250 μm from the surface, while the CF crack initiation sites in the UR12-R and UR12-PR samples occurred at a depth of ~100 μm from the surface (see Fig. 10(b)-(e)). The fatigue crack initiation sites were usually the sensitive locations where the maximal effective tensile stress occurred [61]. Previous investigations indicated that the surface fatigue crack nucleation can be restrained via microstructure refinement [62], compressive residual stress [63], and work hardening [13]. Therefore, the CF crack initiation sites moved towards the subsurface, owing to the combined effect of these three USRP-induced factors, and CRS had a more significant effect. Besides, there was no obvious corrosion in CF crack initiation sites, which indicated that cumulative fatigue damage (mechanical factor) dominated the CF crack initiation of USRPed samples and was completely different from the pitting-induced mechanism in the untreated samples.

Fig. 11 exhibits the surface morphology near the fractography of the investigated samples. The BM sample showed a significant degree of corrosion, as evidenced by deep and large corrosion pits (see Fig. 11(a)). Fig. 11(b) and (d) show that the corrosion pits in the UR12 and UR12-P samples were significantly smaller than those in the BM sample. The UR12 and UR12-P samples experienced prolonged corrosion and a high level of stress. Therefore, the lower pitting degree (compared with that of the BM sample) indicates that the corrosion resistance of the 7B50 aluminum alloy during CF can be efficiently augmented via the 12-pass USRP treatment. Sun et al. [64] have reported a similar result, i.e., the localized corrosion resistance of 7150 aluminum alloy subjected to ultrasonic shot peening was substantially improved due to the surface nanocrystallization and CRS. Owing to the relatively short corrosion fatigue lifetime of the UR12-R and UR12-PR samples, the corrosion pits did not develop to a sufficiently large size on the corresponding sample surfaces, as shown in Fig. 11(c) and (e).

It should be pointed out that the surface corrosion performance analysis of GS layer (see Section 3.2) indicated that the corrosion rate values of USRPed samples in the 3.5% sodium chloride solution were significantly reduced. It was mainly attributed to the introduced compressive residual stress and grain refinement, which were the major beneficial factors. Krawiec et al. [65] reported that laser shock processing led to a considerable increase in the charge transfer and oxide film resistance values, owing to the compressive residual stress generated during the process. Pandey et al. [66] proposed that the surface grain refinement could facilitate the formation of ideal passive film on ultrasonic shot peened 7075 aluminum alloy. Meanwhile, the reaction of aluminum ions and oxygen at the solid-liquid interface was enhanced and the process of aluminum matrix dissolving into the corrosive solution was hindered. The decrease in both corrosion rate and surface pit size (see Fig. 11) of USRP treated samples was helpful in reducing the probability of fatigue crack initiation caused by corrosion defects, thereby improving the corrosion fatigue resistance of the 7B50-T7751 aluminum alloy.

4. Conclusions

Mechanical properties and corrosion fatigue behavior of 7B50-T7751 aluminum alloy treated by 12-pass USRP were investigated in the present study. The major conclusions are summarized as follows:

(1)The formation of a GS layer (thickness: ~425 μm) on the 7B50 alloy was successfully realized. The mean Al grain size is ~ 67 nm at the depth of ~ 40 μm from the top surface. Meanwhile, a 700-μm-deep CRS field is introduced and some micro-cracks initiated in the surface layer.

(2)For GS 7B50 Al alloy, the surface microhardness is increased by 160%, the ultimate tensile strength is slightly improved, but the yield strength, ductility and toughness are slightly reduced. This alloy has a higher strain hardening exponent and is characterized by a higher strain hardening rate, which represents an enhanced resistance to continuous plastic deformation.

(3)In the 3.5% NaCl aqueous solution, the corrosion rate of 7B50 Al alloy following USRP is reduced by 60.08% and its CF strength is increased by 117%. The corrosion resistance and CF life can be further raised after removing the 40-μm-thick surface layer with micro-cracks.

(4)The corrosion rate of the treated alloy was decreased owing to the combined effect of CRS and surface nanocrystallization. The introduced compressive residual stress was the major beneficial factor to improve the corrosion fatigue resistance for the GS 7B50 alloy, and the gradient structure supplied an evident synergistic contribution to this improvement.

Acknowledgments

This work was supported financially by the National Natural Science Foundation of China (No. 51771155) and the Equipment Pre-research Field Foundation (No. 61409220202). The authors are very grateful to Dr. Minghui Ding for her advice on this article.

---