The Effect of Rod-Shaped Long-Period Stacking Ordered Phases Evolution on Corrosion Behavior of Mg95.33Zn2Y2.67 Alloy

The morphology evolution of long-period stacking ordered (LPSO) phases on corrosion behavior of Mg95.33Zn2Y2.67 alloy is investigated systematically during as-cast, pre-extrusion heat-treated, as-extruded and post-extrusion heat-treated conditions. The second phases in the as-cast alloy are only LPSO phases with a few Y particles. The pre-extrusion heat treatment changed LPSO phases from blocks into a rudimentary rod shape with lamellar structure, subsequently into fine fragments by extrusion, and then into a regular rod shape with lamellar structure followed by post-extrusion heat treatment. Immersion tests and electrochemical measurements in 3.5 wt % NaCl solution reveal that the post-extrusion heat-treated alloy has the best corrosion resistance with the lowest corrosion rate. This is attributed to the rod-shaped LPSO phases, which could hinder corrosion proceeding, and result in corrosion orientated along the direction of rods and forming relatively dense long-strip corrosion products. Our findings demonstrate that the improved corrosion resistance of magnesium alloys with LPSO phases can be tailored effectively by the proceeding technology and post-heat treatment.


Introduction
Combined with high specific strength, specific stiffness and excellent casting performance, magnesium and its alloys serve as the lightest structural material, and have become the focus of materials research [1,2]. In particular, magnesium alloy with long-period stacking ordered (LPSO) phases exhibit relatively high strength and good ductility, making them promising candidates for practical applications [3][4][5][6]. Unfortunately, their corresponding corrosion resistance is still a critical factor in restricting their utility. Magnesium alloys with LPSO phases will promote the activity of galvanic microcells, which usually results in a detrimental effect on the corrosion resistance of these alloys [7]. Extensive efforts have been devoted to elucidating the correlation between LPSO phases and corrosion resistance of magnesium alloys [8,9].
It was reported that an 18R LPSO structure formed in Mg-Zn-Y during solidification, while a 14H LPSO structure formed under the constant temperature in Mg-Zn-Gd [10]. R. Schmid-Fetzer et al. [11] confirmed this rule, and further revealed that the existence temperature area of the 18R LPSO structure was 753~483 • C, while the formation temperature of the 14H LPSO structure was lower than 537 • C. The LPSO phase [12] was found to be absent in the as-cast ingots but precipitated during soaking at

Electrochemical Measurements
Electrochemical measurements were tested in 3.5 wt % NaCl solution at room temperature. Specimens were used as working electrodes and mounted in epoxy resin with an exposure area of 0.785 cm 2 (Φ 10 mm). A platinum plate and a saturated calomel electrode were used as the counter and reference electrodes, respectively. Before testing, all the specimens were polished by using SiC sandpaper to 2000# followed by ultrasonic cleaning. Prior to potentiodynamic polarization testing, the open circuit potential of the alloys was measured for 20 min to establish an approximately steady state, followed by polarization curve measurement at a constant scan rate of 1 mV/s. Software CView (CView2, Wuhan CorrTest Co. Ltd., Wuhan, China) was used to fit the polarization curves. The electrochemical impedance spectroscopy (EIS) spectra were tested with the frequency range of 0.1-10 5 Hz and a perturbation amplitude of 10 mV.

Immersion Tests
Samples with dimension of 10 × 10 × 3 mm were polished by SiC sandpaper from 120#-2000# and cleaned ultrasonically for immersion tests in 3.5 wt % NaCl solution at 25 ± 1 • C in an air-conditioned room. The ratio of surface area to solution volume was 1 cm 2 : 20 mL. The evolved hydrogen was collected into a burette above the samples [21]. It was monitored for about 37 h to evaluate the hydrogen evolution rate of the alloys in accordance with [21]. Immersion of 1 h and 98 h was conducted to observe the corrosion characteristics of samples. After 98 h immersion, the samples were removed from the solution, gently rinsed with distilled water, and dried in cold air. Corrosion products were removed using 200 g/L of chromic acid and 10 g/L of AgNO 3 for 10 min. The weights loss were measured to calculate the corrosion rate following the ASTM-G31-72 [22]. The average corrosion rate CR (mm/y) was calculated according to the following equation [22]: where K was a constant 8.76 × 10 4 , W was the weight loss (g), A was the surface area (cm 2 ), T was the time exposed to solution (h), and D was the density of the material (g/cm 3 ). For the immersion tests, three replicates were carried out for each condition to ensure the experimental reproducibility. The corrosion morphologies evolution of the specimens were also investigated by SEM with EDS.

Microstructural Analysis
The XRD result ( Figure 1) of Mg 95.33 Zn 2 Y 2.67 alloys after different processes are mainly composed of Mg and Mg 12 YZn, demonstrating that the second phases in all the current alloys are only LPSO phases. Although the phases are basically the same, the microstructure of these alloys changed significantly. Figure 2 presented a different morphology and distribution of LPSO phases in the alloys of different states: (a) as-cast, (b) pre-extrusion heat-treated, (c) as-extruded, (d) post-extrusion heat-treated. In the as-cast alloy (Figure 2a), the gray phase is the LPSO phase, and the black one is the Mg matrix. LPSO phases (indicated by B) in the as-cast alloys are connected together at the grain boundary in the block. A few white particles (indicated by A) dispersed could be confirmed as Y particles by the EDS results ( Figure 3). After pre-extrusion heat treatment, compared with the as-cast alloys, the LPSO phases at the grain boundary are partially melted and become disconnected, part of which are gathered into a bigger block, like a rudimentary rod shape (indicated by C). In addition, a large number of lamellar LPSO phases (indicated by E) appears in the grain with larger white dots (indicated by D), which were also confirmed as Y particles by EDS (Figure 2b). When the alloy was extruded and crushed, the mass dispersion of different-sized particles are distributed irregularly in the matrix, as shown in Figure 2c. Part of them are extremely deformed, desultorily in a bent-strip shape. After heat treatment at 540 • C for 5 h followed by extrusion, a large number of relative regular, homogeneous rod LPSO phases are formed with a small amount of white dots (Figure 2d). shape. After heat treatment at 540 °C for 5 h followed by extrusion, a large number of relative regular, homogeneous rod LPSO phases are formed with a small amount of white dots (Figure 2d).   shape. After heat treatment at 540 °C for 5 h followed by extrusion, a large number of relative regular, homogeneous rod LPSO phases are formed with a small amount of white dots (Figure 2d).      (Figure 4b), the spacing between two similar lines is 1.61 nm, which also indicates that the LPSO phases obtained in the as-cast Mg-Zn-Y alloy are a 18R structure. As can be seen in Figure 4c, there is an obvious LPSO structure with additional diffraction spots in the LPSO structure with the magnesium matrix [0002], which can be divided into 14 equal parts. High resolution (Figure 4d) shows that the period length of LPSO is 4.1 nm, indicating that the rod-shaped LPSO phase is a 14H-type LPSO phase. In addition, the lamellar phase accompanied by the rod-shaped LPSO phase indicated by E was previously proved to be the same 14H-type LPSO phase as the rod-shaped one. The detailed evolution mechanism of this LPSO structure was also revealed systematically in our previous work [23,24].
In the present study, the alloys included mainly the LPSO phase and α-Mg. The as-cast alloys included 18R LPSO phase and 14H LPSO phase were obtained from the Mg-supersturated solid solution at high annealing temperature. After heat treatment at 540 °C for 5 h, the morphology of the gray block LPSO phase transformed into a regular and uniform gray rod-shaped phase, which is easy to form and difficult to dissolve.   Figure 4a shows a bright field TEM image of the block LPSO phase, which can be easily observed in the SEM image shown in Figure 2a, taken with the incident beam parallel to the [21 10] direction of the hexagonal close-packed (HCP) Mg matrix. The diffraction spots are arranged asymmetrically with respect to the [0001] axis at positions that divide the distance between the incident-beam spot and the [0002] fundamental spot of the simple HCP unit cell into 18 parts. This arrangement indicates that the LPSO phase is made up of an 18-fold periodic stack of close-packed planes. In the high-resolution image (Figure 4b), the spacing between two similar lines is 1.61 nm, which also indicates that the LPSO phases obtained in the as-cast Mg-Zn-Y alloy are a 18R structure. As can be seen in Figure 4c, there is an obvious LPSO structure with additional diffraction spots in the LPSO structure with the magnesium matrix [0002], which can be divided into 14 equal parts. High resolution ( Figure 4d) shows that the period length of LPSO is 4.1 nm, indicating that the rod-shaped LPSO phase is a 14H-type LPSO phase. In addition, the lamellar phase accompanied by the rod-shaped LPSO phase indicated by E was previously proved to be the same 14H-type LPSO phase as the rod-shaped one. The detailed evolution mechanism of this LPSO structure was also revealed systematically in our previous work [23,24].
In the present study, the alloys included mainly the LPSO phase and α-Mg. The as-cast alloys included 18R LPSO phase and 14H LPSO phase were obtained from the Mg-supersturated solid solution at high annealing temperature. After heat treatment at 540 • C for 5 h, the morphology of the gray block LPSO phase transformed into a regular and uniform gray rod-shaped phase, which is easy to form and difficult to dissolve.  Figure 5a shows the polarization curves of Mg95.33Zn2Y2.67 alloys with different LPSO phases in 3.5 wt % NaCl solution. They exhibit non-symmetrical curves and significantly differ from each other. The parameters of corrosion potential Ecorr, corrosion current density icorr, and polarization resistance Rp fitted in Rp mode are listed in Table 1. Compared with the as-cast alloy, both the anode and cathode current density of the pre-extrusion heat-treated alloy are decreased. Namely, heat treatment inhibits the cathode hydrogen evolution and weakens the anode dissolving due to the formation of rod-like LPSO phases. After extrusion, the cathode current density is increased, while the anode current density is decreased, implying the hydrogen evolution during cathode process is enhanced. The reason lies in the fact that fragmentation and deformation of LPSO phases cause a large increase in the number of microcells. The weakened anode reaction is attributed to the more uniform distribution of LPSO phases. It is well known that the uniform distribution of second phases could increase the randomness of galvanic corrosion, and make the corrosion occur uniformly, then decrease the anodic current density [25,26]. Compared with the extruded alloy, the anode current density of the postextrusion heat-treated alloy is not changed, but the cathode current density is decreased. According to the corrosion parameters from Table 1, the rank of corrosion resistance of these alloys is listed as follows: the post-extrusion heat-treated > the pre-extrusion heat-treated > the as-extruded > the ascast. The corrosion resistance of Mg95.33Zn2Y2.67 alloys has been improved to varying degrees after different process. Among them, the post-extrusion heat-treated one has improved the corrosion resistance to the maximum extent. In contrast to other alloys reported previously [9], such as as-cast Mg97Zn1Y2, as-extruded AZ31, as-cast WE43, as-cast ZK60 (Table 1), all the investigated alloys in the present work show a lower corrosion current density. Our previous work proved that the Mg95.33Zn2Y2.67 alloys exhibited good mechanical and damping properties [24]. Currently, it is  Figure 5a shows the polarization curves of Mg 95.33 Zn 2 Y 2.67 alloys with different LPSO phases in 3.5 wt % NaCl solution. They exhibit non-symmetrical curves and significantly differ from each other. The parameters of corrosion potential E corr , corrosion current density i corr , and polarization resistance R p fitted in R p mode are listed in Table 1. Compared with the as-cast alloy, both the anode and cathode current density of the pre-extrusion heat-treated alloy are decreased. Namely, heat treatment inhibits the cathode hydrogen evolution and weakens the anode dissolving due to the formation of rod-like LPSO phases. After extrusion, the cathode current density is increased, while the anode current density is decreased, implying the hydrogen evolution during cathode process is enhanced. The reason lies in the fact that fragmentation and deformation of LPSO phases cause a large increase in the number of microcells. The weakened anode reaction is attributed to the more uniform distribution of LPSO phases. It is well known that the uniform distribution of second phases could increase the randomness of galvanic corrosion, and make the corrosion occur uniformly, then decrease the anodic current density [25,26]. Compared with the extruded alloy, the anode current density of the post-extrusion heat-treated alloy is not changed, but the cathode current density is decreased. According to the corrosion parameters from Table 1, the rank of corrosion resistance of these alloys is listed as follows: the post-extrusion heat-treated > the pre-extrusion heat-treated > the as-extruded > the as-cast. The corrosion resistance of Mg 95.33 Zn 2 Y 2.67 alloys has been improved to varying degrees after different process. Among them, the post-extrusion heat-treated one has improved the corrosion resistance to the maximum extent. In contrast to other alloys reported previously [9], such as as-cast Mg 97 Zn 1 Y 2 , as-extruded AZ31, as-cast WE43, as-cast ZK60 (Table 1), all the investigated alloys in the present work show a lower corrosion current density. Our previous work proved that the Mg 95.33 Zn 2 Y 2.67 alloys exhibited good mechanical and damping properties [24]. Currently, it is revealed that the alloys present considerable anti-corrosion performance, indicating bright prospects for use of the present alloys in light structural materials.     Figure 5b shows the Nyquist curves of the different alloys. It exhibites a capacitive loop at high frequency and a capacitive loop at medium frequency. Only for the extruded alloy, the EIS curve is shown without an inductive arc, indicating that the alloy suffered no localized corrosion. The capacitive loop at high frequency represents the process of charge transfer in a galvanic corrosion, and its radius determines the corrosion resistance of the alloys. At the high-frequency region, the capacitive loop radius of the post-extrusion heat-treated alloy is the largest, suggesting the highest impedance value and the lowest corrosion rate. Meanwhile, with the processing changes, the capacitive loop radius decreases gradually. Moreover, the high-frequency capacitive reactance arcs for the extruded alloy and the as-cast alloy are basically in coincidence, which meant breakage and deformation of LPSO phases has few effects on the charge transfer process and the related corrosion resistance. It is worth mentioning that the distribution of LPSO phases in the extruded state alloy is more uniform, making the alloy more prone to uniform corrosion rather than localized corrosion.

Corrosion Behavior
The high frequency capacitive loops for the pre-extrusion heat-treated and the post-extrusion heat-treated alloys were slightly larger than those of the non-heat-treated alloys. But the diameter of the intermediate frequency capacitive arc was significantly increased. The capacitive loop in the high-frequency region was caused by the reaction between oxide film and corrosion medium. When Mg 2+ reacted with OH − to form Mg(OH) 2 , the barrier effect of the surface layer to the corrosion medium was fortified. However, the Cl − continued to penetrate this oxide layer and reached the metal substrate due to its small volume and strong penetration ability in the corrosive medium. The capacitive loop at the medium frequency region was caused by the corrosion reaction of the metal substrate. Generally, the higher impedance value means the lower the corrosion rate. It was widely accepted that the arc of the intermediate frequency was related to the protective properties of the oxide film formed on the surface of the alloy [25][26][27]. The larger the aperture of the arc, the better the protective performance of the oxide film was, and the more it could inhibit the corrosion reaction. Therefore, the heat treatment could improve the protective properties of the oxide film significantly.
The equivalent circuit models constructed according to the Nyquist plots are shown in Figure 5c,d. Figure 5c was fitted for the as-extruded alloy and Figure 5d was fitted for the other three alloys. The corresponding fitted parameters are listed in Table 2, from which it is easy to find that the post-extrusion heat-treated alloy has the largest corrosion resistance, which verified the conclusions by the polarization curves further.  Figure 6a presents the corrosion behavior of Mg 95.33 Zn 2 Y 2.67 alloys characterized by hydrogen evolution. For each specimen, there is an incubation period with a low rate of hydrogen evolution. After that, the rate of hydrogen evolution is substantially greater, typically tending to a steady state. The incubation period is the longest for the post-extrusion heated alloy and shortest for the as-cast alloy. Thus, the corrosion behavior is reasonably characterized by an initial period of low corrosion rate during an incubation period, followed by a period of increasing corrosion rate, and subsequently steady state corrosion. The overall extent of the corrosion attack is ranked as: the post-extrusion heat-treated < the pre-extrusion heated < the as-extruded < the as-cast. Figure 6b shows the average corrosion rates of Mg 95.33 Zn 2 Y 2.67 alloys under different conditions evaluated by the weight loss method. Among these alloys, the post-extrusion heated one shows the lowest corrosion rate, while the as-cast one shows the highest, which indicates that the corrosion rates of all the Mg 95.33 Zn 2 Y 2.67 alloys are also in the same order as the result of the hydrogen evolution test. Compared to the reported Mg 97.25 Zn 0.25 Y 2 alloy with the LPSO phase [27], the post-extrusion heat-treated alloy of our work presented a much lower corrosion rate, as shown in Figure 6c, indicating the formation of the rod-shaped LPSO phase is an effective way to enhance the corrosion property of Mg-Zn-Y alloys.
work presented a much lower corrosion rate, as shown in Figure 6c, indicating the formation of the rod-shaped LPSO phase is an effective way to enhance the corrosion property of Mg-Zn-Y alloys.

Corroded Surface Evaluation
The morphological surface of as-cast (a) and as-extruded (b) Mg95.33Zn2Y2.67 alloys after EIS tests is shown in Figure 7. Lots of obvious corrosion pits appear on the as-cast one (indicated by the red arrows), while the surface of the as-extruded one is smooth without visible pits. This is because extrusion broke the LPSO phases into uniformly distributed fragments in the alloy, relieving the localized corrosion.

Corroded Surface Evaluation
The morphological surface of as-cast (a) and as-extruded (b) Mg 95.33 Zn 2 Y 2.67 alloys after EIS tests is shown in Figure 7. Lots of obvious corrosion pits appear on the as-cast one (indicated by the red arrows), while the surface of the as-extruded one is smooth without visible pits. This is because extrusion broke the LPSO phases into uniformly distributed fragments in the alloy, relieving the localized corrosion. work presented a much lower corrosion rate, as shown in Figure 6c, indicating the formation of the rod-shaped LPSO phase is an effective way to enhance the corrosion property of Mg-Zn-Y alloys.

Corroded Surface Evaluation
The morphological surface of as-cast (a) and as-extruded (b) Mg95.33Zn2Y2.67 alloys after EIS tests is shown in Figure 7. Lots of obvious corrosion pits appear on the as-cast one (indicated by the red arrows), while the surface of the as-extruded one is smooth without visible pits. This is because extrusion broke the LPSO phases into uniformly distributed fragments in the alloy, relieving the localized corrosion.  (1) as-cast (2)  In order to further clarify the reason why the corrosion resistance of pre-extrusion heat treatment alloys is higher, short-and long-term immersion tests were carried out. As shown in Figure 8, after the short-term immersion of 1 h, a small number of corrosion pits appears on the magnesium matrix, and part of the Y particles are corroded. However, after long-term immersion of 98 h, only the gray LPSO phases existed in the alloy, suggesting that all the Y particles were corroded and peeled off from the Mg matrix. As previously reported [14,28], the cathode activity of the Y particles is more positive than the Mg matrix. The areas around the Y particles are the initiating sites of the galvanic corrosion between the Y particles and the Mg matrix. With the corrosion proceeding, the areas around the Y particles were basically dissolved and the Y particles gradually fell from the matrix. Then the rudimentary rod-shaped LPSO phases were exposed directly in the medium. The appearance of the rod-shaped prototype LPSO phases could act as a physical barrier, preventing the corrosive medium from penetrating into the substrate. Thus, the Mg dissolution reaction is hindered and the corrosion resistance has been improved to some extent. In order to further clarify the reason why the corrosion resistance of pre-extrusion heat treatment alloys is higher, short-and long-term immersion tests were carried out. As shown in Figure 8, after the short-term immersion of 1 h, a small number of corrosion pits appears on the magnesium matrix, and part of the Y particles are corroded. However, after long-term immersion of 98 h, only the gray LPSO phases existed in the alloy, suggesting that all the Y particles were corroded and peeled off from the Mg matrix. As previously reported [14,28], the cathode activity of the Y particles is more positive than the Mg matrix. The areas around the Y particles are the initiating sites of the galvanic corrosion between the Y particles and the Mg matrix. With the corrosion proceeding, the areas around the Y particles were basically dissolved and the Y particles gradually fell from the matrix. Then the rudimentary rod-shaped LPSO phases were exposed directly in the medium. The appearance of the rod-shaped prototype LPSO phases could act as a physical barrier, preventing the corrosive medium from penetrating into the substrate. Thus, the Mg dissolution reaction is hindered and the corrosion resistance has been improved to some extent. The corrosion morphologies of the extruded and post-extrusion heat-treated alloys are compared in Figure 9. After 98 h immersion, the macro-surface of the extruded alloy and the postextrusion heat-treated alloy are apparently different, as shown in Figure 9a,b, respectively. The former is basically corroded with a rough surface, while the latter exhibits a smooth surface with a little corroded area. Furthermore, the corrosion surface for the extruded alloy is covered with plentiful circular-porous products marked with A, which are mainly composed of 70.56 at % O and 29.44 at % Mg by EDS analysis, as shown in Figure 9c. However, the post-extrusion heat-treated alloy surface is covered with very thin and uniform corrosion products, as shown in Figure 9d. From the amplified inset, products like thin strips are distributed out of order. The EDS results indicate the main components are still O and Mg, with a content of 53.76 at % and 46.24 at %, respectively. It is notable that the main compositions of the corrosion products are the same, but the circular products with a high content of O imply a relatively loose and porous structure, and the strip products with a low content of O are compact, dense and uniform. The corrosion morphologies of the extruded and post-extrusion heat-treated alloys are compared in Figure 9. After 98 h immersion, the macro-surface of the extruded alloy and the post-extrusion heat-treated alloy are apparently different, as shown in Figure 9a,b, respectively. The former is basically corroded with a rough surface, while the latter exhibits a smooth surface with a little corroded area. Furthermore, the corrosion surface for the extruded alloy is covered with plentiful circular-porous products marked with A, which are mainly composed of 70.56 at % O and 29.44 at % Mg by EDS analysis, as shown in Figure 9c. However, the post-extrusion heat-treated alloy surface is covered with very thin and uniform corrosion products, as shown in Figure 9d. From the amplified inset, products like thin strips are distributed out of order. The EDS results indicate the main components are still O and Mg, with a content of 53.76 at % and 46.24 at %, respectively. It is notable that the main compositions of the corrosion products are the same, but the circular products with a high content of O imply a relatively loose and porous structure, and the strip products with a low content of O are compact, dense and uniform. The kinetics of magnesium corrosion can be attributed to the properties of the surface film. The formation of Mg(OH)2 film on the magnesium surface is relatively stable compared to other magnesium compounds. However, the existence of strong corrosive ions in aqueous solution, such as Cl − , could severely damage the magnesium surface film. According to the EIS spectrum from Figure 5b, the pore diameter of the intermediate frequency arc of the post-extrusion heat-treated alloy is obviously greater than that of the extruded state alloy, implying the best protective properties of the oxide film formed on the surface. For the post-extrusion heat-treated alloy, the LPSO phases are of different lengths without orientation distribution, as shown in Figure 2d. It was proposed by [28,29] that the LPSO phases were the effective cathode phases in the Mg-Zn-Y alloy, and corrosion generally occurred in the magnesium matrix around the LPSO phases. In this study, since LPSO (e) (f) The kinetics of magnesium corrosion can be attributed to the properties of the surface film. The formation of Mg(OH) 2 film on the magnesium surface is relatively stable compared to other magnesium compounds. However, the existence of strong corrosive ions in aqueous solution, such as Cl − , could severely damage the magnesium surface film. According to the EIS spectrum from Figure 5b, the pore diameter of the intermediate frequency arc of the post-extrusion heat-treated alloy is obviously greater than that of the extruded state alloy, implying the best protective properties of the oxide film formed on the surface. For the post-extrusion heat-treated alloy, the LPSO phases are of different lengths without orientation distribution, as shown in Figure 2d. It was proposed by [28,29] that the LPSO phases were the effective cathode phases in the Mg-Zn-Y alloy, and corrosion generally occurred in the magnesium matrix around the LPSO phases. In this study, since LPSO phases in the post-extrusion heat-treated alloys are of regular and uniform rod shape, corrosion initiated at the adjacent region of the LPSO phases. With the corrosion proceeding, the corrosion products formed along the orientation of rods, resulting in the corrosion products of a long strip rather than being round. Thus, it is easy to understand the extrusion and heat treatment mainly changed the morphology and distribution of the LPSO phases, and then caused the formation of thin-strip corrosion products. These dense and fine strips significantly increased the compactness of the oxide film and enhanced the protective property of the membrane. Furthermore, this rod-shaped LPSO phase is hard to dissolve and could act as a barrier to resist the corrosive ions. Finally, the protective film of corrosion products and the barrier effect of the rod-shaped LPSO phases are combined to influence the corrosion property. Figure 10 shows a schematic map during the corrosion process for Mg 95.33 Zn 2 Y 2.67 alloys under different technical processes. For the as-cast alloy, galvanic corrosion occurs at the Mg matrix around the big LPSO blocks and develops into large and deep holes, which is typical of localized corrosion (Figure 10a). The LPSO blocks turn into a rudimentary rod-shaped structure for the pre-extrusion heated one, which could act as a barrier to hinder the penetration of Cl − (Figure 10b), and improve the corrosion performance to some extent. After extrusion, the LPSO phases are broken into uniformly distributed fragments in the alloy, leading to a more uniform corrosion without localized holes (Figure 10c). That illustrates well why the as-extruded alloy has a higher corrosion resistance than the as-cast alloy. As for the post-extrusion heat-treated one, the rod-shaped LPSO phases lead to the formation of dense and long-strip corrosion products, which are compact to resist the attack of the corrosive ions (Figure 10d). Moreover, the rod-shaped LPSO phases themselves could also act as a barrier to retard the corrosion progress. Thus, these two combined effects enhance the corrosion performance of the post-extrusion heated alloy to a maximum. Our work demonstrates that the improved corrosion resistance of magnesium alloys with LPSO can be tailored effectively by the preceding technology and post-heat treatment. phases in the post-extrusion heat-treated alloys are of regular and uniform rod shape, corrosion initiated at the adjacent region of the LPSO phases. With the corrosion proceeding, the corrosion products formed along the orientation of rods, resulting in the corrosion products of a long strip rather than being round. Thus, it is easy to understand the extrusion and heat treatment mainly changed the morphology and distribution of the LPSO phases, and then caused the formation of thinstrip corrosion products. These dense and fine strips significantly increased the compactness of the oxide film and enhanced the protective property of the membrane. Furthermore, this rod-shaped LPSO phase is hard to dissolve and could act as a barrier to resist the corrosive ions. Finally, the protective film of corrosion products and the barrier effect of the rod-shaped LPSO phases are combined to influence the corrosion property. Figure 10 shows a schematic map during the corrosion process for Mg95.33Zn2Y2.67 alloys under different technical processes. For the as-cast alloy, galvanic corrosion occurs at the Mg matrix around the big LPSO blocks and develops into large and deep holes, which is typical of localized corrosion (Figure 10a). The LPSO blocks turn into a rudimentary rod-shaped structure for the pre-extrusion heated one, which could act as a barrier to hinder the penetration of Cl − (Figure 10b), and improve the corrosion performance to some extent. After extrusion, the LPSO phases are broken into uniformly distributed fragments in the alloy, leading to a more uniform corrosion without localized holes (Figure 10c). That illustrates well why the as-extruded alloy has a higher corrosion resistance than the as-cast alloy. As for the post-extrusion heat-treated one, the rod-shaped LPSO phases lead to the formation of dense and long-strip corrosion products, which are compact to resist the attack of the corrosive ions ( Figure 10d). Moreover, the rod-shaped LPSO phases themselves could also act as a barrier to retard the corrosion progress. Thus, these two combined effects enhance the corrosion performance of the post-extrusion heated alloy to a maximum. Our work demonstrates that the improved corrosion resistance of magnesium alloys with LPSO can be tailored effectively by the preceding technology and post-heat treatment.

Conclusions
The morphological evolution of LPSO phases in the Mg 95.33 Zn 2 Y 2.67 alloys under as-cast, pre-extrusion heat treatment, extrusion, post-extrusion heat treatment can be described as: block LPSO phases with few Y particles → rudimentary rod-shaped LPSO phases with bigger Y phases → fine and deformed fragments of the LPSO phases → regular and uniform rod-shaped LPSO phases with a few Y phases. Their corresponding corrosion resistances are improved to varying degrees and the rank is stated as: post-extrusion heat-treated > pre-extrusion heat-treated > as-extruded > as-cast. Pre-extrusion heat treatment reduced the integral cathode activity and anode dissolution due to the appearance of rudimentary rod-shaped LPSO phases. Extrusion could make the corrosion more uniform because of evenly distributed second phases. The post-extrusion heat treatment caused rod-shaped LPSO phases, which led to the formation of fine-strip corrosion film and prevented the corrosive ions penetrating further effectively. Thus, the corrosion resistance for the post-extrusion heat treatment of Mg 95.33 Zn 2 Y 2.67 alloy has been improved to the greatest extent.
Author Contributions: Jingfeng Wang guide the research ideas of this work and is responsible for the experimental scheme and the submission of the manuscript; Weiyan Jiang is responsible for the research and acquisition, analysis and interpretation of data and manuscript writing. Shengfeng Guo instruct to write and revise the manuscript; Yang Li determined the electrochemical experimental and part of the related analysis; Yao Ma provide some of polishing work and other assistance during the experimental process.