CAF was fabricated by melt-foaming method at the Xinjiang University of China. The amount of thickening agent (high purity Ca about 3 wt% of the melt mass) was added to the molten aluminum (commercial purity Al), and foaming agent (high purity TiH₂, 1.5 wt% of the melt mass) was added. Meanwhile, by the agitation of the foaming agent were dispersed and release gas, after a certain time of heat preservation, the molten aluminum foaming after cooling solidification in the mold was the CAF. The apparent densities of the foams were about 0.67 g∙cm⁻³, cell size was approximately 3 - 7 mm, porosity was 74%, and selecting the same pore structure of CAF specimens was cut into 18 × 11 × 4.5 mm by the electrical discharge machining. SP pellets used in this study were conditioned carbon steel cut wire shot (CCW14) balls with the diameter of ~0.30 mm and the average hardness of pellets was 610 Hv. The specimens were shot-peened on one side with the surface coverage of more than 100%. SP intensity was set at 0.12 mmA, and SP time was 0 s, 5 s, 10 s and 20 s, respectively. Specimens processing status and number were shown in Table 1.

The corrosion resistance of the specimens was obtained by Weight-loss testing [8]. Using total immersion method, the specimens were corroded for 24 hours at the room temperature (20˚C ± 1˚C), with simulate seawater (3.5% NaCl + 1% H₂O₂ in aqueous solution, PH = 5), and then cleaned with 5%HNO₃ + 1%K₂Cr₂O₇ solution. After ultrasonic cleaned in anhydrous ethanol and deionized water washing, the product was dried, and then weighed. The same weight-loss testing experiment was repeated three times to reduce error. The mass loss rate A was calculated according to the following equation:

A = w 0 − w 1 w 0 × 100 % (1)

where w 0 was the initial mass of the specimens before the corrosion and w 1 was the finial mass of the specimen after the corrosion.

Prior to and after the weight-loss testing, the weight of specimen was measured accurately using the SARTORIUS CP225D analytic balance with the accuracy of 10⁻⁴ g. The surface structure analysis of specimens were performed by Bruker D8 Advanced X-ray diffractometer with Cu-K_a radiation (λ = 0.1542 nm), using step-scan ways with 2θ ranging from 30˚ - 120˚. The microstructure was further analyzed by the XRD linear analysis methods. The surface topography was characterized by scanning electron microscope (SEM), and the surface chemical composition of specimens was conducted by the energy dispersive X-ray spectroscopy (EDS).

Figure 1 shows the evolution of the X-ray diffraction patterns with the SP processing

time increasing. It can be seen that the diffraction peaks of Al (PDF#04-0787) was clearly, and there was no other new phase after SP. With the increasing of SP processing time, the intensity of the diffraction peak was significantly weakened and the breath of [111] diffraction peak was broaden obviously. This phenomenon can be explained by the deformation mechanism induced by SP. During the process of SP, a great number of pellets impact the surface of the material from different directions, which causes elastic and plastic deformation at and near surface layer of the specimens [9]. Thus, the domain was refined and the orientation randomization of domain was enhanced, due to the randomization of deformation direction in the surface material after SP.

In order to further obtain the microstructure information of the CAF, the Rietveld whole pattern fitting method was applied with use Materials Analysis Using Diffraction (MAUD) software. The fitting result of the SP3 specimen was showed in Figure 2. It can be seen that, the fitting line and the observed diffraction line were in good agreement, and the weighted profile R_w was 6.7%. The XRD patterns of NP, SP1, and SP2 specimen were also fitted, and the R_w of which were 8.37%, 8.24% and 8.37%, respectively. The R_w values were less than 10%, and which suggested that the results of fitting processing were reliable.

XRD linear analysis method was one of the most effective methods to characterize the microstructure of materials. X-ray diffraction pattern refers to the diffraction intensity with the change curve of the diffraction angle.

d ∗ = 2 sin θ / λ (2)

The observed diffraction linear specimens h ( x ) can be expressed as instrument linear g ( y ) and physical linear f ( x − y ) convolution of specimens. The reason for instrument line broadening were that: 1) The wavelength of the incident X-ray was not completely monochromatic; 2) X-ray light source, specimen, receiving slit was not completely meet the focused condition; 3) The incident

X-ray was not completely parallel etc. The mainly factors affected the physical linear microstructure included: domain size, strain and crystal structural defects responsible for line etc., these factors responsible for line broadening effects was independently.

h ( x ) = ∫ ∞ ∞ g ( y ) f ( x − y ) d y (3)

f ( x − y ) function was related to grain size and strain. Direction-dependent crystallite size and strain models were put forward by Popa [10] [11].

In this model, the normalized peak profile was:

V H ( z ) = ∫ d ( Δ z ′ ) G H d ( Δ z ′ ) L H ( z + Δ z ′ ) (4)

The Gaussian component G_H represents the strain effect, and the Lorentzian component L_H represents the size effect. In terms of the integral breadths β G H , β L H these functions were:

G H ( z ) = β G H − 1 exp ( − π z 2 β G H 2 ) − 1 (5)

L H ( z ) = β L H − 1 ( 1 + π 2 z 2 β L H 2 ) − 1 (6)

For the constant-wavelength diffraction method, z = 2 θ and the quantities β G H β L H were shown as follows.

β G H = 2 tan θ H ( 2 π 〈 ε h k l 2 〉 ) 1 / 2 (7)

β L H = 2 λ ( 3 〈 R h 〉 cos θ H ) (8)

According to the obtained diffraction line parameters and based on the follow principles, direction-dependent domain size and microstrain of the CAF has been obtained from the broadening of diffraction peaks. For face-centered cubic (f, c, c) crystal, The space group was Fm3m, domain size and microstrain expression shown as below:

< D > = D 0 + D 1 K 4 1 ( x , φ ) + D 2 K 6 1 ( x , φ ) + D 3 K 6 1 ( x , φ ) + … (9)

< ε 2 > E H 4 = E 1 ( h 4 + k 4 + l 4 ) + 2 E 2 ( h 2 k 2 + h 2 l 2 + k 2 l 2 ) (10)

The results obtained via XRD linear analysis methods were showed Figures 3-6. It can be seen from Figure 3 that the FWHM increased with the SP time increasing which indicated that the domain was refined and the microstrain was increased after SP.

Figure 4 shows the domain size varies with the SP time. It was clear that the domain of major peaks (111) and (200) was refined obviously after 5 s-SP, and then only a slight variation can be discerned with the SP time prolonged. The results reveal that 5S was the optimal processing time.

The variation trend of microstrain was shown in Figure 5, which was inversed of Figure 4. An interesting phenomenon can be found that, the microstrain decreased firstly and then increased with the SP time prolonged. This phenomenon may be due to CAF special porosity structure and property. During the SP processing,

the deformation of cell structure and shear collapse of CAF was random, and the magnitude of the stress was not affected.

The average domain size along with SP time was characterized by the lognormal distribution model, and results were shown in Figure 6. As the processing time increased, the average domain sizes decreased and domain sizes lognormal distribution curve obviously narrowed [12].

The effect of SP treatment on corrosion resistance of CAF was also discussed. It can be seen from Figure 7 that the mass loss rate of the shot-peened specimens was smaller than that of the unpeened specimen. It should be clear that SP can effectively improve the corrosion resistance. Compared with the SP1, SP2 and SP3 specimens, the mass loss rate of SP1 was the smallest, which indicated that the corrosion resistance of SP1 was best. This experiment results showed that 5 seconds was the optimal processing time, which was consistent with the Rietveld fitting results. The surface nanocrystallization has been induced, and composition was more homogeneous than unpeened specimens [13]. The difference of electrochemical decreased in grain interior and grain boundary, and local corrosion on interface diminished. Meanwhile, the residual compressive stresses (RCS) on the surface would inhibit corrosion crack initiation and propagation [14]. And the ability of corrosion resistance was improved.

The surface morphologies were observed by SEM from Figures 8(a)-(d). Figure 8(c) showed the CAF cell local area appear perforated hole, this phenomenon made pore connectivity increased, the materials and corrosive medium real contact area increased, the corrosion resistant decreased. Meanwhile, Figure 8(d) showed that microcracks were induced by SP .This phenomenon suggests with the processing time increasing, the damage and failure of the surface materials were serious. The existence of perforated hole and microcrack decreased the corrosion resistance of CAF. In all, effect the surface nanocrystallization and

compressive residual stress field still remain dominate, which were much better than disadvantage brought by the surface damage of the CAF.

The chemical composition of specimen’s surface was analyzed by X-ray spectroscopy (EDS). The element contents were given in Table 2. It can be seen from that the major compound of surface materials was aluminium oxide. Compared with atomic percentage of NP and SP1, which of the elements O increased so that more aluminium oxides were generated. It was meaning that large fraction of the non-equilibrium grain boundaries furnished the specimens by SP, with

more nucleus to form denser nature oxide film than ordinary grain [15]. This denser oxide film could be keeping more stable and relative integral in an oxidative atmosphere. Meanwhile, compared with atomic percentage of the SP1, SP1 and corrosion, the elements O also increased. It obviously showed that the percentage of the elements O in the specimens surface SP1 and corrosion was higher than the NP and SP1. It was meaning that corrosion can cause the stress relaxation, and the deep oxidation occurred more easily.