The Corrosion Behaviors of an As-Rolled Mg-8Li (in wt.%) Alloy in Two Differently Concentrated NaCl Solutions

: By means of an electrochemical workstation, hydrogen evolution device, optical microscope (OM) and scanning electron microscope (SEM), the corrosion behaviors of an as-rolled Mg-8%Li alloy with a dual phase structure in 0.9 wt.% NaCl and 3.5 wt.% NaCl solutions have been investigated and compared. The results show that when the immersion time exceeds 8 h, the hydrogen evolution rate of the alloy in the 0.9 wt.% NaCl is 3 times higher than that in the 3.5 wt.% NaCl solution. Moreover, the corrosion behaviors of the alloy are obviously different in the two differently concentrated NaCl solutions. In the 3.5 wt.% NaCl solution, the localized corrosion is much more severe and can occur simultaneously in the interior of both the α -Mg and β -Li matrix phases. However, the localized corrosion in the 0.9 wt.% NaCl solution is obviously weak and mainly occurs at the α -Mg phase.


Introduction
Magnesium-lithium (Mg-Li) alloys are the lightest metallic materials known and their density is between 1.35~1.65 g/cm 3 , which is only 1/2 of that of aluminum alloys. Compared with traditional magnesium alloys, Mg-Li alloys have great application prospects in the fields of automotive, aerospace and the military due to their excellent deformation capacity, specific strength and stiffness [1][2][3][4][5]. Generally, the microstructure of Mg-Li alloys varies with the added Li content. The dual phase Mg-Li alloys is comprised of the α-Mg phase and β-Li phase when the Li content is less than 10.3 wt.% and higher than 5.5 wt.% [5]. It has been widely reported that the duplex structured Mg-Li alloys exhibit a good combination of mechanical strength and plasticity [6][7][8][9][10][11][12], ensuring their wide application prospects compared to other Mg alloys. However, micro galvanic corrosion can easily occur between these two phases because the electrode potential of the β-Li phase is lower than that of the α-Mg phase, resulting in the poor corrosion resistance of dual phase Mg-Li alloys [3,[13][14][15][16][17].
Since the chloride ion (Cl − ) is an aggressive ion and has strong activity, the corrosion behavior of magnesium alloys is generally reported to be positively correlated with chloride ion concentration [18][19][20]. It has been reported that the corrosion severity of uncoated and coated AZ31 samples was significantly increased with the concentration of chloride ions [20]. In the research of the corrosion behavior of AZ31D magnesium alloys in NaCl solutions, Shan et al. reported that due to the increased concentration of chloride ions, the high frequency capacitive loop shrank, and the open circuit potential shifted to a more negative direction [19]. However, it is also reported that the induced pitting corrosion in the simulated acid rain solutions was not sensitive to the chloride ion concentrations, but due to the AlMn phases, it was located at the grain boundary [21]. Accordingly, it can be concluded that the corrosion behavior of Mg alloys in NaCl solutions depends on the Cl − concentration and the microstructural characteristics. For Mg-Li based alloys, although

Sample Preparation
The experimental material used in this work is a typical (α-Mg + β-Li) dual phase Mg-8%Li (in wt.%) alloy. Through the inductively coupled plasma atomic emission spectrum (ICP-AES) apparatus, its chemical composition was determined, as listed in Table 1. The alloy was prepared by melting the high purity Mg and Li bulks in a vacuum reaction furnace under the atmosphere of flux argon to avoid burning. After keeping for 40 min at 710 • C to homogenize it, the molten alloy was cast into plate ingots with a cross-section of 200 mm × 300 mm and a thickness of 50 mm. Then, thin plates with a cross-section of 150 mm × 150 mm and a thickness of 18 mm were cut from the ingot. After being homogenized at 300 • C for 30 min, the plates were rolled at room temperature and their thickness was controlled to be 6 mm. To eliminate the residual stress induced by the rolling process, plates were held at 300 • C for 1 h in an air furnace. Then, block samples with a dimension of 10 mm × 5 mm × 5 mm for electrochemical measurements were cut from the plates and mounted by the epoxy resin AB glue with an exposed working area of 5 mm × 10 mm. Then, samples were grinded gradually with 400#, 800#, 1000# and 2000# water abrasive papers to remove the surface scratches, followed by washing with acetone and drying with cold air. Moreover, samples for the immersion and hydrogen evolution tests were polished with 2.5 µm diamond paste after grinding. To clearly reveal the microstructure of the alloy, the polished surface of one sample was etched by an enchant solution of (4% HNO 3 + 96% C 2 H 5 OH) and subsequently observed by using an optical microscope (OM). Based on the "ImageJ" software, the occupied area fractions of two matrix phases were measured.

Electrochemical Tests
The CS350 electrochemical workstation (Corrtest Co. Ltd., Wuhan, China) was used for the electrochemical test. The traditional three-electrode set-up was applied, i.e., the exposed areas of the mounted samples were the working electrode, the saturated calomel electrode (SCE) was the reference electrode and the platinum plate was the auxiliary electrode. After being pre-immersed in 0.9 wt.% NaCl and 3.5 wt.% NaCl solutions, respectively, for 0, 4, 8 and 24 h, their electrochemical measurements are performed. Before the measurements of potentiodynamic polarization curves and electrochemical impedance spectrum, a delay of 10 min was required to obtain a stable open circuit potential. The scanning rate of potentiodynamic polarization measurements is 0.166 mv/s. Potentiodynamic polarization curves were fitted by CorrView software. Electrochemical impedance spectra (EIS) were measured in a frequency range from 100 kHz to 10 mHz with a disturbed potential amplitude of 10 mV and subsequently fitted by using the ZSimDemo 3.30 software. To reveal the deviation of the measured data, each electrochemical experiment was carried out at least three times. Finally, the electrochemical curves were drawn by Origin 2019 software.

Hydrogen Evolution Measurements and Immersion Test
It has been reported that when Mg alloys are exposed to the aqueous solutions, the hydrogen evolution reaction due to the cathodic reaction (i.e., 2H + + 2e → H 2 ↑) could take place [32,33]. Thus, the measured volume of the evolved hydrogen is related to the anodic dissolution of the substrate of Mg alloys [32]. The schematic diagram of the hydrogen evolution device used in the current investigation is shown in Figure 1. The polished samples were respectively immersed into the 0.9 wt.% NaCl and 3.5 wt.% NaCl solutions. The ratio of the corrosion medium volume to the exposed area of samples was set to 200:1. The measurements of hydrogen evolution volume were performed up to 48 h and the data were recorded every 2 h. After being respectively immersed in 0.9 wt.% NaCl and 3.5 wt.% NaCl solution for 4, 8 and 24 h, the surface morphologies and their threedimensional (3D) profiles were observed by using a stereo optical microscope (VHX-900F, Keyence International, Mechelen, Belgium). The operational principle for obtaining 3D images with precise changes in height includes two steps: (1) taking the image at the same location automatically layer by layer; (2) combining all images into one. Moreover, the cross-sectional morphologies of the immersed samples were observed by using a scanning electron microscope (SEM; FEI Quanta 450, Hillsboro, OR, USA).  Figure 2 shows the optical microstructure of the as-rolled Mg-8%Li alloy. It can be seen that the alloy exhibits a typical dual phase structure. Generally, for the Mg-Li binary alloys, they are composed of α-Mg and β-Li phases when the added Li content varies from 5.5 to 10.3 wt.% [3,5,10,14]. Moreover, for the duplex structured Mg-Li alloys, the α-Mg and β-Li matrix phases respectively appear in white and black when observed by using OM [3,5,10,14]. Based on the Mg-Li binary phase diagram proposed by Masing et al. [34],    Figure 2 shows the optical microstructure of the as-rolled Mg-8%Li alloy. It can be seen that the alloy exhibits a typical dual phase structure. Generally, for the Mg-Li binary alloys, they are composed of α-Mg and β-Li phases when the added Li content varies from 5.5 to 10.3 wt.% [3,5,10,14]. Moreover, for the duplex structured Mg-Li alloys, the α-Mg and β-Li matrix phases respectively appear in white and black when observed by using OM [3,5,10,14]. Based on the Mg-Li binary phase diagram proposed by Masing et al. [34], the start melting temperature of β-Li phase is lower than that of the α-Mg phase and decreases from 592 • C to 180.6 • C with the increase of Li content in the β-Li phase. Therefore, the α-Mg phase first nucleates and grows at random sites during solidification process, resulting in their irregular and messy distribution. On the basis of the measurement from Image J software, the area fraction of the α-Mg phase is determined to be 60%.  Figure 2 shows the optical microstructure of the as-rolled Mg-8%Li alloy. It can be seen that the alloy exhibits a typical dual phase structure. Generally, for the Mg-Li binary alloys, they are composed of α-Mg and β-Li phases when the added Li content varies from 5.5 to 10.3 wt.% [3,5,10,14]. Moreover, for the duplex structured Mg-Li alloys, the α-Mg and β-Li matrix phases respectively appear in white and black when observed by using OM [3,5,10,14]. Based on the Mg-Li binary phase diagram proposed by Masing et al. [34], the start melting temperature of β-Li phase is lower than that of the α-Mg phase and decreases from 592 °C to 180.6 °C with the increase of Li content in the β-Li phase. Therefore, the α-Mg phase first nucleates and grows at random sites during solidification process, resulting in their irregular and messy distribution. On the basis of the measurement from Image J software, the area fraction of the α-Mg phase is determined to be 60%.   Figure 3 shows the potentiodynamic polarization curves of the samples being respectively pre-immersed in the two NaCl solutions for 0, 4, 8 and 24 h. It can be clearly seen that their cathodic branches are basically overlapped, indicating that the Cl − concentration has almost no impact on the cathodic process. Since the negative difference effect and anodic dissolution occur during anodic polarization and then have a great impact on the sample surface, the cathodic and anodic branches of the polarization curves are asymmetric and the current density in the anodic section increases rapidly. As a result, the cathodic branch is generally used for fitting analysis [28]. The Tafel fitted results of potentiodynamic polarization curves are listed in Table 2. This reveals that the corrosion current densities of the samples being pre-immersed in the 0.9 wt.% NaCl are less than those being pre-immersed in the 3.5 wt.% NaCl. In addition, the corrosion potentials of the samples are much higher in the 0.9 wt.% NaCl than those in the 3.5 wt.% NaCl. After performing pre-immersion for up to 24 h, the corrosion current density of the sample in the 0.9 wt.% NaCl increased from the initial value of 1.9735 × 10 −5 to 5.104 × 10 −5 A/cm 2 . However, the current density of the sample in the 3.5 wt.% NaCl corrosion medium increased from the initial value of 2.1679 × 10 −5 to 6.1598 × 10 −5 A/cm 2 . Thus, the corrosion current density of samples in two concentrated solutions increases continuously with the prolongation of pre-immersion time. Moreover, the increased rate of corrosion current density of the samples due to the pre-immersion in the 3.5 wt.% NaCl is much higher than those in the Coatings 2022, 12, 406 5 of 14 0.9 wt.% NaCl. Generally, the lower corrosion current density ensures the alloys have better corrosion resistance [35]. Compared with the corrosion in the 3.5 wt.% NaCl solution, samples immersed in the 0.9 wt.% NaCl have better corrosion resistance.

Electrochemical Responses
NaCl increased from the initial value of 1.9735 × 10 to 5.104 × 10 A/cm . Howe current density of the sample in the 3.5 wt.% NaCl corrosion medium increased f initial value of 2.1679 × 10 −5 to 6.1598 × 10 −5 A/cm 2 . Thus, the corrosion current de samples in two concentrated solutions increases continuously with the prolong pre-immersion time. Moreover, the increased rate of corrosion current density of t ples due to the pre-immersion in the 3.5 wt.% NaCl is much higher than those in wt.% NaCl. Generally, the lower corrosion current density ensures the alloys hav corrosion resistance [35]. Compared with the corrosion in the 3.5 wt.% NaCl s samples immersed in the 0.9 wt.% NaCl have better corrosion resistance.    Figure 4 shows the AC impedance spectrum of the sample immersed in the 0.9 wt.% NaCl and 3.5 wt.% NaCl solutions for 0, 4, 8 and 24 h. For the samples measured in two concentrations of NaCl solutions, a high-frequency capacitive loop and a middle to low-frequency capacitive loop can be observed (Figure 4a). Moreover, the radius of the capacitive loop of the sample immersed in 3.5 wt.% NaCl is smaller than that in 0.9 wt.% NaCl. Generally, the larger the radius of the capacitive loop, the better the corrosion resistance of the sample [3,14,27,30]. Thus, the corrosion resistance of the sample immersed in 0.9 wt.% NaCl is much better. After being pre-immersed for 4 h, the impedance spectrum of the sample in 0.9 wt.% NaCl shows a high-frequency capacitive loop and a middle to low-frequency capacitive loop, whilst it contains a high frequency capacitive and a low-frequency inductive loop for the sample in the 3.5 wt.% NaCl solution (Figure 4b).
Generally, the appearance of the low-frequency inductive loop indicates that the sample surface is prone to pitting or localized corrosion [21]. Therefore, it indirectly reflects that the occurrence of pitting or localized corrosion may happen at the surface of the sample being immersed in the 3.5 wt.% NaCl solution for 4 h. After being pre-immersed for 8 h, the sample immersed in 0.9 wt.% NaCl contains a high-frequency capacitive loop, a middle to low-frequency capacitive loop and a low-frequency inductive loop, whilst the sample immersed in 3.5 wt.% NaCl contains a high-frequency capacitive and a lowfrequency inductive loop (Figure 4c). Moreover, the radius of each of the capacitive loop of the two curves is reduced, demonstrating that the corrosion resistance decreases with increasing immersion time in both concentrated NaCl solutions. After being pre-immersed for 24 h, the AC impedance spectrums of the samples measured in two concentrated NaCl solutions contain a high frequency capacitive and a low-frequency inductive loop (Figure 4d). Moreover, the radius of their capacitive loops is further reduced. For the sample immersed in the 0.9 wt.% NaCl solution, the middle to low-frequency capacitive loop disappears. Based on the changes of the capacitive loop radius, this reflects that the corrosion resistance of the samples in the two NaCl solutions decreases further with the prolongation of immersion time. Additionally, the localized corrosion attack may occur earlier in the 3.5 wt.% NaCl solution, indicating that the corrosion resistance of the sample is much better in the 0.9 wt.% NaCl solution.  To reflect the corrosion characteristics of the alloy being pre-immersed in two NaCl solutions for different times, electrochemical equivalent circuits were proposed to fit the EIS data, as shown in Figure 5. In the equivalent circuits, Rs is the solution resistance. Rct and Qdl represent charge transfer resistance and electric double layer, respectively, at the interface between substrate and electrolyte, which are used to describe the capacitance To reflect the corrosion characteristics of the alloy being pre-immersed in two NaCl solutions for different times, electrochemical equivalent circuits were proposed to fit the EIS data, as shown in Figure 5. In the equivalent circuits, R s is the solution resistance. R ct and Q dl represent charge transfer resistance and electric double layer, respectively, at the interface between substrate and electrolyte, which are used to describe the capacitance loop at high frequency. Q dl is a constant phase element that replaces an ideal capacitor to account for the non-homogeneity in the system, which is defined by the two values of Y dl and n dl . Among them, n dl is the dispersion coefficient of Q dl . If n dl = 1, Q dl is identical to a capacitor; if n dl = 0, Q dl represents a resistance. R f and Q f (defined by Y f and n f ) represent film resistance and capacity in the medium frequency capacitance loop, respectively. R L and L stand for resistance and inductance, respectively, which are used to describe the low frequency inductance loop and related to the localized corrosion [36].  The fitted results are listed in Table 3. It reveals that for the samples measured differently concentrated NaCl solutions, their Rct values decrease with the prolon of pre-immersed time. Moreover, the Rct values measured in the 3.5 wt.% NaCl so are obviously lower than those in the 0.9 wt.% NaCl solution. Thus, the corros sistance of the Mg-8%Li alloy in the 0.9 wt.% NaCl solution is relatively higher a creases with the prolongation of pre-immersed time. In addition, the low frequen ductance loop in the 3.5 wt.% NaCl solution occurs earlier than that in the 0.9 wt.% solution, indicating that the occurrence of localized corrosion in the 3.5 wt.% NaC tion is much easier.  The fitted results are listed in Table 3. It reveals that for the samples measured in two differently concentrated NaCl solutions, their R ct values decrease with the prolongation of pre-immersed time. Moreover, the R ct values measured in the 3.5 wt.% NaCl solution are obviously lower than those in the 0.9 wt.% NaCl solution. Thus, the corrosion resistance of the Mg-8%Li alloy in the 0.9 wt.% NaCl solution is relatively higher and decreases with the prolongation of pre-immersed time. In addition, the low frequency inductance loop in the 3.5 wt.% NaCl solution occurs earlier than that in the 0.9 wt.% NaCl solution, indicating that the occurrence of localized corrosion in the 3.5 wt.% NaCl solution is much easier.

Hydrogen Evolution
Since the cathodic process of electrochemical corrosion of Mg alloys is due to the hydrogen evolution reaction, their corrosion rate can be calculated by the volume fraction of evolved hydrogen per unit of exposed area. Figure 6 shows the hydrogen evolution curves of the samples immersed in two concentrated NaCl solutions for up to 48 h. It reveals that after immersion for 48 h, the hydrogen evolution volumes of the samples immersed in 0.9 wt.% and 3.5 wt.% NaCl solutions are 1.16 and 2.69 mL/cm 2 , respectively, indicating that the corrosion rate of the sample in the 3.5 wt.% NaCl solution is faster than that in the 0.9 wt.% solution. This result is consistent with the electrochemical measurements. Moreover, the variations of hydrogen evolution rates of the samples measured in the two concentrated solutions are different during the immersion process. When the immersion time is less than 8 h, the hydrogen evolution rates of the samples measured in the two concentrated solutions are low and their difference is quite small. When the immersion time exceeds 8 h, the hydrogen evolution rates of the samples immersed in the two concentrated solutions increase remarkably. Based on the slopes of hydrogen evolution curves, the hydrogen evolution rate of the alloy in the 3.5 wt.% NaCl solution is 3 times higher than that in the 0.9 wt.% NaCl solution.

Hydrogen Evolution
Since the cathodic process of electrochemical corrosion of Mg alloys is due to the hydrogen evolution reaction, their corrosion rate can be calculated by the volume fraction of evolved hydrogen per unit of exposed area. Figure 6 shows the hydrogen evolution curves of the samples immersed in two concentrated NaCl solutions for up to 48 h. It reveals that after immersion for 48 h, the hydrogen evolution volumes of the samples immersed in 0.9 wt.% and 3.5 wt.% NaCl solutions are 1.16 and 2.69 mL/cm 2 , respectively, indicating that the corrosion rate of the sample in the 3.5 wt.% NaCl solution is faster than that in the 0.9 wt.% solution. This result is consistent with the electrochemical measurements. Moreover, the variations of hydrogen evolution rates of the samples measured in the two concentrated solutions are different during the immersion process. When the immersion time is less than 8 h, the hydrogen evolution rates of the samples measured in the two concentrated solutions are low and their difference is quite small. When the immersion time exceeds 8 h, the hydrogen evolution rates of the samples immersed in the two concentrated solutions increase remarkably. Based on the slopes of hydrogen evolution curves, the hydrogen evolution rate of the alloy in the 3.5 wt.% NaCl solution is 3 times higher than that in the 0.9 wt.% NaCl solution.

Observations to the Corrosion Morphologies
To reveal the different corrosion mechanisms of samples immersed in the two concentrated NaCl solutions, their surface corrosion morphologies and 3D images were observed and compared, as shown in Figures 7 and 8, respectively. At the early stage of corrosion, obvious filiform-like corrosion occurs on the surfaces of the different samples. Moreover, the corrosion severity of the samples in the two solutions intensifies with the prolongation of immersion time. When the immersion time is 4 h, the corrosion severity of the sample immersed in the 0.9 wt.% NaCl solution is significantly slighter than that in the 3.5 wt.% NaCl solution (Figures 7a and 8a). Based on the measured results from 3D images, the maximum depths of corrosion pits in the 0.9 wt.% and 3.5 wt.% NaCl solutions are 6.65 µm and 14.07 µm, respectively. Since a dense layer of Li 2 CO 3 can be formed on the surface of the β-Li phase [3,5,[29][30][31], the subsequent corrosion attack to the substrate can be effectively inhibited. For the sample that is immersed in the 3.5 wt.% NaCl solution, the corrosion attack is much more severe and occurs widely in the α-Mg phase. Moreover, the corrosion severity increases, and the depths of the corrosion pits become deeper when the immersion time increases. It is well known that the head of the filiform tip has the stronger acidity and can accelerate the corrosion, whilst the back of the filiform tip is alkaline and corrosion products are deposited [3]. When the immersion time is up to 24 h, some β-Li phases can still be present for the sample immersed in the 0.9 wt.% NaCl solution, whilst the whole surface of the sample that is immersed in the 3.5 wt.% NaCl solution is almost completely covered by the corrosion products. However, the maximum depths of the corrosion pits in the two solutions after immersion for 24 h are basically same and their values are 20.55 µm and 21.56 µm, respectively. Combined with surface corrosion morphology and 3D images, it can be concluded that the corrosion attack of the Mg-8%Li alloy is much more severe in the 3.5 wt.% NaCl solution, which is consistent with the results of the hydrogen evolution and electrochemical tests.   Figure 9 shows the cross-sectional corrosion morphologies of the samples that are immersed in the two concentrated NaCl solutions for 4 h, respectively. It reveals that the localized corrosion attack of the sample immersed in the 3.5 wt.% NaCl is much more severe than that in the 0.9 wt.% NaCl (Figure 9a,b). Based on the high magnification images (Figure 9c), there are almost no visible corrosion pits on the surface. However, for the sample in 3.5 wt.% NaCl, the maximum depth of the corrosion pits can reach 10 μm (Figure 9b,d). Compared with the surface and cross-sectional observations, this indicates that although the filiform corrosion can easily occur on the surface of the Mg-8%Li alloy in differently concentrated NaCl solutions, the localized pitting is quite sensitive to the concentration of Cl − . High magnification images clearly show that the pitting can only  Figure 9 shows the cross-sectional corrosion morphologies of the samples that are immersed in the two concentrated NaCl solutions for 4 h, respectively. It reveals that the localized corrosion attack of the sample immersed in the 3.5 wt.% NaCl is much more severe than that in the 0.9 wt.% NaCl (Figure 9a,b). Based on the high magnification images (Figure 9c), there are almost no visible corrosion pits on the surface. However, for the sample in 3.5 wt.% NaCl, the maximum depth of the corrosion pits can reach 10 µm (Figure 9b,d). Compared with the surface and cross-sectional observations, this indicates that although the filiform corrosion can easily occur on the surface of the Mg-8%Li alloy in differently concentrated NaCl solutions, the localized pitting is quite sensitive to the concentration of Cl − . High magnification images clearly show that the pitting can only occur in the 3.5 wt.% NaCl solution and preferentially nucleates in the interior of the α-Mg matrix phases. Moreover, with the intensification of pitting corrosion, the electrolyte can reach underneath matrix and cause the dissolution of the β-Li phases because the Li 2 CO 3 film is hardly formed due to the limited concentration of CO 2 in the solution, resulting in the corrosion of the whole surface of the sample that is immersed in the 3.5 wt.% NaCl solution for 24 h (Figure 8e). In a previous work, Song et al. reported that when the duplex structured Mg-8 wt.%Li alloy immersed in NaCl solution, the formed surface film was composed of MgO, Li 2 O, Mg(OH) 2 and LiOH [37]. Moreover, due to the reaction between Li 2 O and CO 2 , the Li 2 CO 3 can be formed on the surface of β-Li matrix phase. Generally, the Pilling-Bedworth ratio (PBR) of the formed film is between 1 and 2 [29]. It can have good protective capability for the underneath substrate. Zeng [29]. In the previous work, Xu et al. reported that the formed Li 2 CO 3 film was dense and could protect the corrosion attack of the β-Li matrix phase, whereas the MgCO 3 and Mg(OH) 2 films formed on the surface of α-Mg phase were porous and loose [30]. Moreover, the Cl − can easily destroy the MgCO 3 and Mg(OH) 2 films formed on the surface of the α-Mg phase [3]. Since the Cl − concentration in the 3.5 wt.% NaCl solution is high, the stronger electrical conductivity of the solution can accelerate the electrochemical reactions and increase the corrosion rate of the alloy. Moreover, the high concentration of Cl − can cause much more severe attacks to the surface films and reduce their corrosion protectiveness to the substrate. Thus, the difference of corrosion resistance of the alloys in the two concentrated NaCl solutions reaches the maximum when the immersion time is less than 4 h. With the increasing of the immersion time, the quantity of formed corrosion products on the sample surfaces will increase considerably. Then, the corrosion resistance due to the Cl − attack can be offset by the increased corrosion products on the surfaces. This explains why the difference of corrosion resistance between the samples immersed in the 0.9 wt.% and 3.5 wt.% NaCl solutions decreases with the prolongation of immersion time. When the immersion time reaches 24 h, the corrosion severity on the surfaces of the differently immersed samples in the 0.9 wt.% and 3.5 NaCl solutions are basically the same (Figures 7f and 8f). Moreover, the initially localized corrosion increases the exposed area of the sample surfaces. Thus, when the immersion time exceeds 8 h, the hydrogen evolution rates of the samples immersed in the two concentrated NaCl solutions increase slightly.

Conclusions
Through investigating and comparing the corrosion behavior of an as-rolled Mg-8%Li alloy in differently concentrated NaCl solutions, two conclusions can be drawn: (1) Based on the determined slopes of the measured hydrogen evolution curves, the corrosion rate of the alloy in the 3.5 wt.% NaCl solution is 3 times higher than that in the 0.9 wt.% NaCl solution when the immersion time exceeds 8 h. (2) The corrosion mechanisms of the as-rolled Mg-8%Li alloy are obviously different in the two concentrated NaCl solutions. In the 0.9 wt.% NaCl solution, the corrosion attack mainly occurs at the α-Mg phase. In the 3.5 wt.% NaCl solution, the corrosion attack initially occurs in the α-Mg phase and then gradually extends to the β-Li phase.

Conclusions
Through investigating and comparing the corrosion behavior of an as-rolled Mg-8%Li alloy in differently concentrated NaCl solutions, two conclusions can be drawn: (1) Based on the determined slopes of the measured hydrogen evolution curves, the corrosion rate of the alloy in the 3.5 wt.% NaCl solution is 3 times higher than that in the 0.9 wt.% NaCl solution when the immersion time exceeds 8 h. (2) The corrosion mechanisms of the as-rolled Mg-8%Li alloy are obviously different in the two concentrated NaCl solutions. In the 0.9 wt.% NaCl solution, the corrosion attack mainly occurs at the α-Mg phase. In the 3.5 wt.% NaCl solution, the corrosion attack initially occurs in the α-Mg phase and then gradually extends to the β-Li phase.