1 Introduction

As the lightest structural material, magnesium alloys not only have high specific strength and specific stiffness but also have good castability, machinability, thermal conductivity, electromagnetic shielding ability and excellent damping characteristic, which makes magnesium alloys promising candidate materials in the fields of aerospace, automobile and electronics [1,2,3,4]. However, the application of magnesium alloys is limited due to their relatively low absolute strength, insufficient ductility, and poor corrosion resistance. It is reported that adding an appropriate amount of rare earth (RE) elements into magnesium alloys is an effective way to improve the comprehensive properties of magnesium alloys [5, 6]. Of all RE-containing magnesium alloys, Mg–RE–Zn systems have attracted extensive attention due to their excellent mechanical properties and creep resistance [7,8,9]. Mg–RE–Zn alloys can be divided into two types according to whether there is long period stacking ordered (LPSO) phase in as-cast state [10,11,12,13]. For TypeIMg–RE–Zn alloys, LPSO phase forms at grain boundaries during the casting process (RE = Y, Dy, Ho, Er, and Tm). For TypeIIMg–RE–Zn alloys, LPSO does not form during the casting process but forms in the subsequent annealing process (RE = Gd and Tb).

Gadolinium (Gd) has very high solubility in magnesium alloys (23.5 wt% at the eutectic temperature) and, therefore, the Mg–Zn–Gd alloy system is an important alloy system among RE-containing magnesium alloys. Liu et al. [14] found that, with the increase of Gd content in Mg–Gd–Zn alloys, the alloy microstructure was gradually refined and the compression performance at room temperature was gradually enhanced. Wang et al. [15] found that addition of Gd to magnesium alloys would inhibit dynamic recrystallization and increase texture intensity, leading to improved mechanical properties; the Mg–Gd–Zn alloy with 2 wt% Gd exhibited a yield strength of 369.8 MPa and ultimate tensile strength of 400.3 MPa. Similarly, Sabbagian et al. [16] found that, by adding 1 wt% Gd to the Mg–Zn alloy system, the average grain size was reduced from 14.6 to 5.7 μm; the yield strength was increased from 198 to 241 MPa; and the ultimate tensile strength was increased from 301 to 336 MPa.

Compared with the Mg–Zn–Gd alloy system, the Mg–Zn–Y alloy system has attracted more attention due to the presence of LPSO phase in the as-cast state [17]. It is reported that the Mg–Zn–Y alloy prepared by rapid powder metallurgy process shows a yield strength of 600 MPa and an elongation of 5% at room temperature due to the presence of fine LPSO phase at grain boundaries [18, 19]. The LPSO phase could enhance the alloy strength through short-fiber strengthening mechanism [20, 21]. Further, the LPSO phase would accommodate the stress concentration and coordinate plastic deformation of the alloy, leading to improved plasticity before cracking.

In addition to mechanical properties, corrosion resistance of RE-containing magnesium alloys is another important issue [22, 23]. It is generally believed that the micro-galvanic corrosion between the second phase and the magnesium matrix is a key factor affecting the corrosion resistance of the magnesium alloy [24,25,26,27,28]. Some studies show that the addition of Gd and Y can improve the microstructure of magnesium alloys and weaken the micro-galvanic corrosion. For instance, Zhang et al. [29] found that LPSO phase in the Mg–Zn–Y alloy was more difficult to corroded than the matrix, so it could work as a barrier between the matrix and eutectic phase. Zhang et al.[30] elucidated that minor Gd addition in MgZn1Y2 (at.%) could refine the grain size and form complete network eutectic and large volume fraction of LPSO phase, resulting in enhanced corrosion resistance. However, the detrimental effect of RE-containing phases on corrosion resistance of magnesium alloys has also been frequently reported. Srinivasan et al.[31] studied the corrosion behavior of Mg–Gd–Zn alloy in NaCl solution and found that the relatively discrete network structure became continuous with the increase of Gd content. At the same time, with the increase of the volume of the second phases ((Mg, Zn)3Gd and LPSO), serious micro-galvanic corrosion occurred and the corrosion rate increased. Liu et al.[32] observed that the accelerated corrosion rate of the cast Mg–2Zn–15Gd-0.39 Zr alloy was mainly attributed to the presence of micro-cathodes, such as (Mg, Zn)3Gd eutectic phase and the LPSO phase.

The literature review suggests that the effect of Gd and Y on the corrosion behavior of RE-containing magnesium alloys varies with alloying element content, the type and distribution of the second phase. This is probably why there are many arguments regarding the effect of Gd and Y on the corrosion behavior of RE-containing magnesium alloys. Therefore, in present study, the corrosion behavior of the Mg–Zn–Y and Mg–Zn–Gd alloys with the same element content (at.%) was studied comparatively. We found that the corrosion rate of the MgZn2Y2.66 alloy, in terms of the corrosion current density, was about one order of magnitude lower than that of the MgZn2Gd2.66 alloy. The high corrosion resistance of the MgZn2Y2.66 alloy was mainly attributed to the presence of LPSO phase in form of continuous networks and the relatively high corrosion resistance of the corrosion product layer on the alloy.

2 Experimental

2.1 Material Preparations

The MgZn2Y2.66 and MgZn2Gd2.66 alloys were obtained by melting pure Mg (99.9 wt%), Mg-26.38 wt% Gd, Mg-21.5 wt% Y and pure Zn in a resistance furnace with CO2 and SF6 protection gas. Firstly, the surface of the master alloys was fully polished to remove the oxide film. Then the polished alloy, pure Mg and pure Zn were put into a stainless steel crucible and then heated in a resistance furnace. The cast alloy was obtained by keeping the melt at 750 ℃ for 25 min and then cooled in a saltwater bath. Table 1 lists the actual chemical composition of these alloys. Samples for microstructural characterization, electrochemical measurements and immersion tests were prepared from the ingots. Each sample was sequentially grinded with sandpapers from 320# to 3000#, ultrasonically cleaned in distilled water, and then dried in a cool air stream.

Table 1 Chemical composition of the alloys
Full size table

2.2 Microstructural Characterization

Samples for microstructural observation were first mechanically ground using SiC sandpapers and then electrochemically polished in commercial AC2 solution for 2 min at 20 V. The microstructure of the alloys was examined in a Zeiss Sigma HD field emission gun scanning electron microscope (FEG-SEM) equipped with Oxford energy dispersive spectroscopy (EDS). The alloy phases in the alloys were analyzed using PANalytical-Empyrean X-ray diffraction (XRD), with a copper target, with a scanning angle range of 20°–90° and a scanning speed of 2°/min.

2.3 Immersion Test

The immersion test was carried out in 3.5 wt% NaCl solution at room temperature. During the immersion process, a stereo microscope (SMZ, SZN71) was used to observe and record the macroscopic corrosion morphology of the sample surfaces. Further, the hydrogen evolution test and mass loss test were performed on parallel samples immersed in 3.5 wt% NaCl solution at room temperature for 24 h. After immersion, the corrosion products were collected. Their chemical composition and chemical state were analyzed by an X-ray photoelectron spectroscope (XPS, Thermofisher, ESCALAB, 250Xi) with an Al Kα X-ray radiation source. The surface and cross-sectional morphology of selected samples were examined by FEG-SEM. The cross sections of the samples were prepared by ultramicrotomy (Leica Ultracut). The corrosion products of selected samples were removed by immersing the samples in a mixed solution containing 200 g/L chromic acid and 10 g/L of AgNO3 according to the national standard GB/T16545-1996.

2.4 Electrochemical Measurements

Electrochemical measurements were carried out on a Gamry Reference 3000 potentiostat at room temperature in 3.5 wt% NaCl solution. The sample, a platinum plate and a saturated calomel electrode were used as the working electrode, counter electrode and reference electrode, respectively. Potentiodynamic polarization and electrochemical impedance spectroscopy tests were conducted after immersion in the solution for 300 s to obtain a relatively stable surface condition.

3 Results

3.1 Microstructure Observation

Figure 1 shows backscattered electron micrographs of the as-cast alloys, where the alloy matrix was revealed as dark areas while the second phases were revealed as bright areas. The morphology and distribution of second phases were distinctly different in the two alloys. The major second phase in the Mg–Zn–Y alloy had a lamellar structure and was distributed in a network along the grain boundary (Fig. 1a, b). In contrast, the major second phase in the Mg–Zn–Gd alloy was in fishbone-like form (Fig. 1c, d) and was distributed discontinuously along grain boundaries. Table 2 lists the EDS analysis results of points A, B, C and D in Fig. 1, corresponding to the major second phase and solid solution in the two alloys, respectively. Combining with the EDS analysis, the XRD analysis (Fig. 2) and literature reports, it can be inferred that the lamellar phase in Mg–Zn–Y alloy is Mg12YZn (LPSO) phase [33] and the fishbone-like phase in Mg–Zn–Gd alloy is (Mg, Zn)3Gd (W phase) [34, 35].

Fig. 1
figure 1

Backscattered SEM images of as-cast MgZn2Y2.66 alloy a, b and MgZn2Gd2.66 alloy c, d

Full size image
Table 2 EDS elemental analysis of the phase marked in Fig. 1
Full size table
Fig. 2
figure 2

XRD patterns of alloy: a MgZn2Y2.66, b MgZn2Gd2.66

Full size image

3.2 Corrosion Behavior

3.2.1 Immersion and Hydrogen Test

Figure 3 shows surface morphology of the alloy recorded in situ when the alloys were immersed in 3.5 wt% NaCl solution. At the initial stage of immersion (~ 3 min), many bubbles appear on the surface of both alloys, and there are fewer bubbles on Mg–Zn–Y alloy (Fig. 3a) than Mg–Zn–Gd alloy (Fig. 3b). With the increase of immersion time, the number of bubbles on the alloy surface decreases and the size increases gradually (~ 15 min). After that, the number of bubbles on the alloy surface further reduced and eventually disappeared. At the same time, localized corrosion characterized by stable gas streams appeared on the alloy surfaces. It can be known from the principle of corrosion electrochemistry that the bubbles and gas streams formed during the immersion process are related to hydrogen gas formed at the cathode when the magnesium alloy dissolves at the anode.

Fig. 3
figure 3

Macroscopic corrosion morphologies of the two alloys in 3.5 wt% NaCl solution: a MgZn2Y2.66 alloy, b MgZn2Gd2.66 alloy

Full size image

The uniformly-distributed gas bubbles on the alloy surface in the early state of immersion suggest uniform corrosion at the moments, and the corrosion rate is low. Thereafter, with the formation and thickening of the corrosion product layer, the uniform corrosion rate of the alloy decreased. Consequently, the number of the gas bubbles on the surface also decreased, and some small bubbles merged and grew into larger bubbles. When the immersion process was further extended, the corrosion form changed from uniform corrosion to localized corrosion. At this time, the corrosion concentrated in local parts of the alloy, which resulted in anodic dissolution of these parts and cathodic protection of the surrounding regions.

Figure 4a, b shows the volume of hydrogen gas vs. immersion time of the alloys recorded during immersion in the NaCl solution for up to 24 h. The volume of the hydrogen gas was small for both alloys in the first 2 h of immersion, suggesting relatively low corrosion rate. After that, the volume of the hydrogen gas increased with immersion time for both alloys. However, the increasing rates are different. For the MgZn2Y2.66, the volume of the hydrogen gas increased at a relatively low rate during the whole course of the immersion. In contrast, the volume of the hydrogen gas for the MgZn2Gd2.66 increased quickly between 2 and 8 h of immersion, suggesting fast propagation of corrosion attack in the alloy. Beyond that, the volume of the hydrogen gas increased at a slightly-reduced rate, suggesting slightly reduced corrosion rate. As discussed later, this is probably related to the formation of the corrosion product layer. Figure 4c shows the mass loss of the alloys after immersion in the NaCl solution for 24 h. The total mass loss of the MgZn2Gd2.66 alloy is about 6 times that of the MgZn2Y2.66 alloy, suggesting much higher corrosion rate of the MgZn2Gd2.66 alloy compared with the MgZn2Y2.66 alloy.

Fig. 4
figure 4

a Average hydrogen evolution of studied alloys in 3.5 wt% NaCl solution, b the enlargement of the red rectangle region in a; and c mass loss rate of studied alloys in 3.5 wt% NaCl solution

Full size image

3.2.2 Electrochemical Test

In order to assist the understanding of the corrosion mechanism of the alloys, electrochemical measurements were carried out. Figure 5a shows the open circuit potentials (OCPs) of the two alloys recorded during immersion in the NaCl solution for up to 300 min. The OCP of the Mg–Zn–Y alloy increased rapidly from − 1.72 to − 1.54 V and then decreased to − 1.60 V in the initial stages of immersion (0–20 min). Thereafter, the OCP fluctuated around − 1.60 V with the extension of the immersion time. In contrast, the OCP of the Mg–Zn–Gd alloy increased at reduced rates from − 1.61 to − 1.54 V within 60 min and then stabilized. The increase of OCP at the initial stage indicates that a corrosion product layer has been formed on the alloy surfaces. The increasing rate of the OCP reflects the increasing resistance of the corrosion product layer which is determined by the composition, compactness, and thickness of the corrosion product layer. The relatively higher stabilized OCP of the Mg–Zn–Gd alloy relative to the Mg–Zn–Y alloy is probably associated with the relatively more positive standard electrode potential of Gd/Gd+3 (− 2.28 V) compared with that of Y/Y+3 (− 2.37 V) [36].

Fig. 5
figure 5

a Open circuit potential and b potentiodynamic polarization curves of the two alloys in 3.5 wt% NaCl solution

Full size image

Figure 5b shows potentiodynamic polarization curves of the two alloys in the NaCl solution. Compared with Mg–Zn–Gd alloy, the polarization curve of Mg–Zn–Y alloy was significantly shifted to the left, suggesting higher corrosion resistance of the Mg–Zn–Y alloy relative to the Mg–Zn–Gd alloy. The corrosion potentials and corrosion current densities of the two alloys obtained by Tafel fitting of the date are listed in Table 3. The corrosion current density of the Mg–Zn–Y alloy was one order of magnitude lower than that of the Mg–Zn–Gd alloy.

Table 3 Corrosion potentials and corrosion current densities of the two alloys obtained by Tafel fitting of the potentiodynamic polarization curves
Full size table

The electrochemical impedance spectra of the two alloys after immersion in 3.5 wt% NaCl solution for 5 min are shown in Fig. 6. The Nyquist plots of the MgZn2Y2.66 alloy displayed one capacitive loop at high and intermediate frequencies, while the Nyquist plots of MgZn2Gd2.66 alloy displayed one capacitive loop at high and intermediate frequencies and an inductive loop at low frequencies. In the Bode plots, the MgZn2Y2.66 alloy shows relatively high impedance at low frequencies, suggesting formation of a continuous corrosion product layer; while the impedance of the MgZn2Gd2.66 alloy dropped dramatically at low frequencies, suggesting initiation and propagation of localized corrosion. The equivalent circuit model used to fit the electrochemical impedance spectra of the two alloys is shown in Fig. 7. Rs is the solution resistance; Rct refers to the charge transfer resistance; Qdl refers to the double electric layer capacitance at the interface between matrix and solution (capacitance ring in the high-frequency region). Rf and Qf are film resistor and capacitor (capacitor rings in the mid-frequency region), respectively. RL and L stand for resistance and inductance, respectively, being used to describe low-frequency inductive circuits or the onset of localized corrosion. In the current study, proper fitting is carried out on Zsimpwin software, and the estimated results are listed in Table 4. One can see that Rct of MgZn2Y2.66 is much higher than MgZn2Gd2.66, suggesting that MgZn2Y2.66 alloy has relatively higher corrosion resistance. This is consistent with the immersion test and the potentiodynamic polarization test. Furthermore, the Rf value MgZn2Y2.66 is much higher than MgZn2Gd2.66, suggesting that the corrosion product layer on the MgZn2Y2.66 alloy has higher resistance than that on the MgZn2Gd2.66 alloy. The result well agrees with the OCP analysis.

Fig. 6
figure 6

Electrochemical impedance spectra (EIS) of the two alloys after immersion in 3.5 wt% NaCl solution for 5 min: a Nyquist diagram, b Bode diagram

Full size image
Fig. 7
figure 7

Equivalent circuit used to fit the EIS shown in Fig. 6

Full size image
Table 4 Parameters obtained by fitting the EIS shown in Fig. 6 using the equivalent circuit as shown in Fig. 7
Full size table

3.2.3 Corrosion Products Analysis

XPS was conducted to determine the composition of the collected corrosion products. The XPS survey spectra reveal the existence of Mg, O, and Y elements in MgZn2Y2.66 alloy (Fig. 8a), and Mg, O, and Gd elements in MgZn2Gd2.66 alloy (Fig. 8b). The Mg 1s peak in the high-resolution spectra can be divided into two sub-peaks: Mg(OH)2 and MgO. Based on the XPS results, the corrosion products are mainly composed of MgO/Mg(OH)2 and Y2O3 in the MgZn2Y2.66 alloy (Fig. 8c, e), while the corrosion products consist of Gd2O3, Gd(OH)2, MgO/Mg(OH)2 in the MgZn2Gd2.66 alloy (Fig. 8d, f).

Fig. 8
figure 8

XPS survey spectra of the corrosion products: a MgZn2Y2.66, b MgZn2Gd2.66. High-resolution XPS for c Mg 1s and e Y 3d of MgZn2Y2.66, d Mg 1s and f Gd 4d of MgZn2Gd2.66

Full size image

3.2.4 Corrosion Morphology

The surface morphology of the two alloys after immersion for 300 min (MgZn2Y2.66) and 170 min (MgZn2Gd2.66), respectively, in the 3.5 wt% NaCl solution were examined to understand the corrosion mechanism (the corrosion products were removed), as shown in Fig. 9. It can be seen from Fig. 9a, c that there are regularly arranged Y and Zn-rich elements on the alloy surface, and the surface distribution patterns of Y and Zn are similar. According to Fig. 1, many LPSO phase remains on the surface of the Mg–Zn–Y alloy after corrosion, indicating that the corrosion rate of the LPSO is lower than that of the magnesium matrix on the one hand, and the corrosion degree of the whole alloy is not severe on the other hand. It can be seen from Fig. 9b, d that there are randomly arranged Gd and Zn-rich substances on the alloy surface, and the surface distribution of the two elements is quite different. According to Fig. 1, the residual substances on the surface of the Mg–Zn–Gd alloy are related to the W phase. These W phase-related Gd and Zn-rich materials are randomly distributed and have different shapes. This suggests that although W has a lower corrosion rate than the Mg matrix, the W phase itself has also undergone certain corrosion damage during the immersion process. In addition, the corrosion attack of alloy is severe. Perhaps, the W phase at different depths was accumulated after corrosion, leading to the distribution pattern of Gd and Zn-rich substances different from that of the original W phase.

Fig. 9
figure 9

Microscopic corrosion morphologies of the two alloys after immersion in 3.5 wt% NaCl solution: a, c MgZn2Y2.66 for 300 min, b, d MgZn2Gd2.66 for 170 min

Full size image

In order to further understand the corrosion mechanism of the two alloys, the cross sections of the alloys after immersion in 3.5 wt% NaCl solution for short term and long term were observed (without removing corrosion products). The cross sections morphology of the alloys after immersion for 5 min was shown in Fig. 10. On the surfaces of MgZn2Y2.66 alloy, a corrosion products layer of ~ 520 nm thickness was formed, which mainly resulted from uniform corrosion of the Mg matrix, as shown in Fig. 10a. In contrast, the corrosion product layer on the surface of MgZn2Gd2.66 alloy is about 200 nm thickness, which is much thinner than that on the surface of MgZn2Y2.66 alloy. Furthermore, the corrosion products on the surfaces of the LPSO phase (indicated by the orange arrow) are very thin and there is a localized corrosion site within the LPSO phase (indicated by the red arrow). This suggests that although the LPSO phase is cathodic to the Mg matrix, the Mg-rich layers within the LPSO phase are also susceptible to corrosion attack.

Fig. 10
figure 10

Cross-sectional morphologies of the alloys after immersion in 3.5 wt% NaCl solution for 5 min: a MgZn2Y2.66, b MgZn2Gd2.66

Full size image

Figure 11 shows the cross-sectional morphology of the MgZn2Y2.66 alloy after immersion for 5 h. As can be seen from Fig. 11a, although immersion time is as much as 5 h, the corrosion attack of the alloy mainly concentrated on the surface, and the corrosion front mostly stopped at the Mg/LPSO phase interface. In Fig. 11b, c, the continuous network of LPSO acts as an obstacle in the corrosion propagation path, which effectively prevents corrosion from propagating along the Mg matrix into the depth. Interestingly, Fig. 11d shows that corrosion can spread in certain directions within the LPSO phase. It is worth pointing out that the interior of LPSO phase is composed of Zn/Y layers and Mg-rich layers. Therefore, it is inferred that the Mg-rich layers in LPSO phase would also undergo anodic dissolution, just like the Mg matrix.

Fig. 11
figure 11

a Low magnification cross-sectional morphology of the MgZn2Y2.66 alloys after immersion in 3.5 wt% NaCl solution for 5 h; b–d the high-magnification micrographs of blue, orange and green rectangle regions in a, respectively

Full size image

In contrast to the MgZn2Y2.66 alloy, severe corrosion attack occurred on the surface of the MgZn2Gd2.66 alloy only after immersion for 3 h, and there was no obvious relationship between the corrosion front and the distribution of W phase (Fig. 12a). Observations at higher magnifications are shown in Fig. 12b–d. It can be clearly seen from Fig. 12a–d that the W phase is more resistant to corrosion than the Mg matrix, but the W phase will still be destroyed as the corrosion progresses. Further, it is clearly shown in Fig. 12c, d that a tunnel passing through the W phase and connecting two neighboring grains was developed, suggesting that the W phase network failed to stop corrosion propagation.

Fig. 12
figure 12

Low magnification a and high magnification b cross-sectional morphologies of the MgZn2Gd2.66 alloys after immersion in 3.5 wt% NaCl solution for 3 h; c the high-magnification micrograph of blue rectangle region in a, d the high-magnification micrograph of orange rectangle region in b

Full size image

4 Discussion

It has been demonstrated that the MgZn2Y2.66 alloy is much more corrosion resistant than the MgZn2Gd2.66 alloy, but the mechanism for this phenomenon is still unknown and deserves further discussion. In early stages of immersion, both alloys mainly suffered from uniform corrosion. In this case, the corrosion resistance of the alloy can be indirectly reflected by the characteristics of the corrosion product layer formed after uniform corrosion. According to Table 2, the Mg matrix of the two alloys contains similar amount of alloying elements, with the major difference in the type of RE element. The XPS analysis (Fig. 8) indicates that, in addition to MgO/Mg(OH)2, the MgZn2Y2.66 alloy has Y2O3 in its corrosion product layer while the MgZn2Gd2.66 alloy has Gd2O3 and Gd(OH)2. Such compositional difference in corrosion product layer should be, to some degree, responsible for the different responses of the two alloys during electrochemical measurement. Further, Fig. 10 shows that the corrosion product layer on MgZn2Y2.66 alloy is thicker than that on the MgZn2Gd2.66 alloy. This is probably associated with the different phase particles in the two alloys.

It has been reported [37] that the potential difference between the LPSO phase and the Mg matrix is 250 mV in Mg-Y-Zn alloy, while the potential difference between W phase and the Mg matrix is 120 mV in Mg–Zn–Gd-Zr alloy [38]. Therefore, the LPSO phase in MgZn2Y2.66 alloy would trigger more severe galvanic corrosion than the W phase in the MgZn2Gd2.66 alloy. In the initial stage of immersion, the LPSO phase in the MgZn2Y2.66 alloy would hardly be corroded. Uniform corrosion of the Mg matrix led to the formation of a relatively thick corrosion product layer. As for the MgZn2Gd2.66 alloy, the uniform corrosion rate is relatively low and corrosion product layer is relatively thin, due to relatively weaker galvanic coupling effect between the W phase and Mg matrix.

With the extension of immersion time, the corrosion gradually changed from uniform corrosion to localized corrosion, as seen in Fig. 3. According to the surface morphology of the removed corrosion products, it can be seen that although LPSO and W phases are both cathodes, they will experience different degrees of corrosion, and LPSO phase is more corrosion resistant than W phase. More importantly, LPSO phase has a strong corrosion barrier effect, which can effectively prevent propagation of localized corrosion from one grain to surrounding grains, as shown in Fig. 11. Since localized corrosion could also occur within the LPSO phase, it is possible that the electrolyte can penetrate the LPSO block though the Mg-rich layers within the LPSO phase, leading to corrosion attack of beneath alloy matrix. However, as reported previously, the volume fraction of the LPSO phase in the MgZn2Y2.66 alloy is close to 50%, which can guarantee a very low corrosion propagation rate, even after immersion in the NaCl solution for 24 h [39]. In contrast, the W phase has almost no blocking effect on the propagation of localized corrosion, as suggested in Fig. 12. Besides, once localized corrosion propagates into the alloy matrix, an occluded micro-cell will be formed, further accelerating corrosion attack of the alloy. This explains why the corrosion rate of the MgZn2Gd2.66 alloy is much higher than that of the MgZn2Y2.66 alloy, particularly between 2 and 8 h of immersion in the NaCl solution. In the late stages of immersion, e.g., 8 h or longer, the MgZn2Gd2.66 alloy was so severely corroded that many corrosion products deposited on the alloy surface, which might slow down the mass exchange between the electrolyte and the alloy, resulting in mild reduction of the corrosion rate.

5 Conclusions

In the present study, the effect of yttrium (Y) and gadolinium (Gd) on the corrosion behavior of RE-containing magnesium alloys was systematically studied. Particular attention was focused on the difference in composition of the corrosion products and the role of second phases in the two alloys. The main findings are as follows:

  1. 1.

    The microstructure of as-cast MgZn2Y2.66 and MgZn2Gd2.66 alloys is mainly composed of Mg matrix and the second phases. The second phase in the MgZn2Y2.66 alloy is the Mg12YZn (LPSO) phase, while the second phase in the MgZn2Gd2.66 alloy is the (Mg,Gd)3Zn (W) phase.

  2. 2.

    The corrosion behavior of the two alloys is similar when immersed in 3.5%NaCl solution. In early stages, uniform corrosion dominated the corrosion process and the corrosion rate is relatively low. After exposed to the NaCl for about 15 min, localized corrosion gradually became dominated and the corrosion rate started to increase. However, corrosion rate of the MgZn2Gd2.66 alloy is much higher than that of the MgZn2Y2.66 alloy after prologed exposure, e.g., 60 min.

  3. 3.

    In early stages, the corrosion product layer on the MgZn2Y2.66 alloy surface is relatively thick and composed of MgO/Mg(OH)2 and Y2O3, while the corrosion products on the MgZn2Gd2.66 alloy are relatively thin and consists of Gd2O3, Gd(OH)2 and MgO/Mg(OH)2. This is one reason for the high corrosion resistance of the MgZn2Y2.66 alloy. After prolonged exposure, the network of LPSO phase in the MgZn2Y2.66 alloy can effectively prevent corrosion from propagating within Mg matrix while the W phase in MgZn2Gd2.66 does not have such effect, which is another reason for the high corrosion resistance of the MgZn2Y2.66 alloy.