Effect of Manganese on Microstructure and Corrosion Behavior of the Mg-3Al Alloys

Abstract

Microstructure and corrosion behavior of the Mg-3Al-xMn (x = 0, 0.12, 0.21, 0.36, 0.45) (hereafter in wt.%) alloys were experimentally investigated by electron probe microanalysis (EPMA), scanning electron microscope equipped with energy dispersive X-ray spectroscopy (SEM/EDX), X-ray diffraction (XRD), electrochemical, and hydrogen evolution tests. A new self-constructed Mg-Al-Mn-Fe thermodynamic database was used to predict the solidification paths of the alloys. The addition of Mn showed no grain refinement in the cast Mg-3Al alloys. According to the microstructure observation, Al-Fe phases were observed in the non-Mn-added alloy, while Al₈Mn₅(LT) (Al₈Mn₅ in low temperature) became the main intermetallic phase in the Mn-added alloys, and the amount increased gradually with the Mn addition. The τ–Al_0.89Mn_1.11 phase with lower Al/(Fe + Mn) ratio was observed in the alloys with 0.36 and 0.45 wt.% Mn content. According to the electrochemical tests, all five alloys showed localized corrosion characteristics in 3.5 wt.% NaCl solution. Compared with the Mg-3Al alloy, the corrosion resistance of Mn-added alloys were significantly improved and increased gradually with the Mn addition, which was due to the variation of Al-containing intermetallic compounds. The present experimental investigations and thermodynamic calculations confirmed the mechanism that the increasing amount of Al₈Mn₅(LT) with Mn addition could encapsulate the B2-Al(Mn,Fe) phase with higher Fe. Therefore, it could prevent this detrimental phase from contacting magnesium matrix, thus suppressing micro-galvanic corrosion and improving corrosion resistance gradually.

1. Introduction

Magnesium alloys attract a lot of attention due to weight-sensitive applications such as automotive, aerospace, and electronics industries [1,2,3,4,5]. Corrosion performance is one of the major concerns for applications of magnesium alloys [6,7]. Influences of alloy elements such as Ce [8,9], Nd [9,10,11,12,13], Zn [14,15], Y [11,16], Ca [11,17], and Mn [9,11,15,18,19,20,21,22,23] on the corrosion behavior of magnesium alloys have been investigated widely in literature. It is widely admitted that Mn addition could combat impure Fe in Mg-Al alloys and reduce corrosion rates of the Mg-Al alloys [11,18,19]. It is reported that the Fe/Mn ratio should not exceed a threshold value. If it goes beyond the ratio, the corrosion rates would increase significantly [24]. For AM60, the ratio is 0.021, which means this alloy could tolerate Fe impurity up to 0.021 of the Mn concentration (by weight) [25,26,27,28,29,30]. This ratio should be related to the phase constitution of the alloys. Intermetallic compounds such as Al₈Mn₅(LT) [31,32,33,34,35,36,37,38,39,40], τ–Al_0.89Mn_1.11 [31,41], Al₁₁Mn₄(LT) (Al₁₁Mn₄ in low temperature) [38], and Al₄Mn [42,43] have been reported in Mg alloys. Previous studies reported the correlation between amount of elements and corrosion behavior. Using calculated phase diagrams and Scheil solidification, Liu et al. [28] reported that the detrimental effect of Fe on the Mg corrosion was caused by the precipitation of the Fe-rich Bcc_B2 phase. Recently, Metalnikov et al. [19] experimentally investigated the influence of large Mn addition (0.5, 1.4, 3.1 wt.%) on microstructure and corrosion behavior of the wrought Mg-5Al alloys. The Mg-5Al-0.5Mn alloy was reported to have the best corrosion performance among the alloys in their results, and Al₈Mn₅(LT) and (βMn) were identified in the Mn-added alloys. The formation of the Al-Mn(-Fe) intermetallic phases with Mn addition needs systematical investigation to gain a better understanding of the mechanisms of improving corrosion resistance in Mg-Al alloys with Mn-addition. Thermodynamic calculations based on CALPHAD (calculation of phase diagram) type thermodynamic databases have proved to be effective in precipitation and corrosion of Mg alloys [28,44]. It is undisputed that the accuracy of the thermodynamic database is primarily important for the related calculations and predictions [45,46,47,48]. Al [49] and Mn [50] both were reported to improve the mechanical properties of Mg alloys. Mn and Fe have an opposite effect on the corrosion properties [51]. Fe has extremely deleterious effects on corrosion resistance due to the ability to serve as an active cathodic site [24], while Mn addition could combat this effect by forming less active Al-Mn-(Fe) particles [51]. Therefore, the Mg-Al-Mn-Fe system is of great interest, and it is worth further investigating via the thermodynamic calculations. In the past few years, the thermodynamic descriptions of some sub-binary and ternary systems of the Mg-Al-Mn-Fe system have been updated [52,53,54,55,56,57].

Therefore, this work combines microstructure observations, thermodynamic calculations, and corrosion examinations with three aims: (i) to experimentally investigate the effect of Mn addition on the microstructure of the Mg-3Al alloys, focusing on the identification of Al-containing intermetallic compounds and evolution of morphologies and chemical compositions; (ii) to investigate the influence of Mn on corrosion behavior of the Mg-3Al alloys; and (iii) to explain some microstructure observations and corrosion behavior with thermodynamic calculations based on a new self-constructed Mg-Al-Mn-Fe thermodynamic database.

2. Materials and Methods

2.1. Materials

In total, five Mg-3Al-xMn (x = 0, 0.12, 0.21, 0.36, 0.45) alloys were prepared via the approach described below. Magnesium ingots (99.95 wt.%) were heated up to be melted at 690 °C under the protection of an argon gas mixture containing SF₆ and N₂ in a resistance furnace. Subsequently, a certain amount of aluminum ingots (99.97 wt.%) and Mg-5Mn master alloy were added into the Mg bath, and the temperature was maintained at 690 °C for 120 min. Then, the melts were poured into a steel mold and cooled with the furnace. The chemical compositions of the alloys were determined by ICP-AES (inductively coupled plasma-atomic emission spectroscopy, ICAP 7000 type, Thermo Fisher Scientific, Waltham, MA, USA), as shown in

Table 1. Alloy composition (wt.%) determined by ICP-AES (inductively coupled plasma-atomic emission spectroscopy).

Alloys	Al	Mn	Fe	Cu	Ni	Si	Fe/Mn	Mg
Mg-3Al	2.97	0.01	0.013	0.001	0.006	0.001	1.300	Balance
Mg-3Al-0.12Mn	2.98	0.12	0.010	0.001	0.015	0.001	0.083	Balance
Mg-3Al-0.21Mn	2.86	0.21	0.010	0.001	0.030	0.001	0.048	Balance
Mg-3Al-0.36Mn	2.97	0.36	0.008	0.003	0.019	0.001	0.022	Balance
Mg-3Al-0.45Mn	2.96	0.45	0.015	0.002	0.018	0.001	0.033	Balance

. It should be mentioned that traces of Mn in the Mg-3Al alloy resulted from the impurity in raw materials. The samples were sectioned from the central part of the ingots and cut into 10 mm × 10 mm × 10 mm cubic by wire-cut machine (DK7720 type, Fangzheng CNC Machine Tool Co., Ltd., Taizhou, China). In order to eliminate the disturbance of the Al₁₂Mg₁₇ phase, the samples were held at 400 °C for 15 h in an air furnace, followed by quenching in cold water. To remove the oxidation film, the surface of samples after heat treatment were ground off before experimental tests.

2.2. Experimental Methods

In order to identify phase type, phase constitution, and morphologies of the Al-containing intermetallic particles in the Mg-3Al-xMn alloys, XRD (D8 Advance type, Bruker Company, Stuttgart, Switzerland), EPMA (JXA-8530F type, JEOL, Tokyo, Japan) and SEM/EDX (Helios NanoLab G3 UC type, FEI Company, Hillsboro, OR, USA) were performed. To identify the type of intermetallic particles, we adopted a particles extraction method [31,40,41] described below. The different alloys were immersed in acetic acid aqueous solution (9 vol.%). After the sample was dissolved, the compound powders in the solution were centrifuged, washed with pure water thrice and with ethanol twice, and then collected after being dried at 50 °C for 5 h. Phase identification of these particles was conducted by XRD. The patterns were indexed by the software Jade (Version 6, MDI, Livermore, CA, USA). The composition of the intermetallic particles was determined by EPMA, because WDS (wavelength dispersive X-ray spectroscopy, JEOL, Tokyo, Japan) is more accurate when performing quantitative analysis compared with EDX. The samples for EPMA tests were ground with SiC papers progressively up to 2000 grit, mechanically polished by 0.5 μm diamond paste, and ultrasonically cleaned with ethanol. To reveal the grain boundaries, the samples for EPMA were electrolytic polished. The mechanical polished samples were connected with the anode of the power supply and a stainless steel bar with the cathode. After being polished under 15 V direct current for about 30 s in a mixture solution (10 mL perchloric acid and 190 mL ethanol), the samples were ultrasonically cleaned in ethanol and dried in the air. In addition, SEM/EDX was applied to obtain the high magnification morphology and the elemental distribution of the intermetallic particles. The samples for SEM/EDX were mechanically polished and then etched by nitric acid ethanol solution (4 vol.%) for 10 min to reveal the 3-dimensional (3D) morphologies of the intermetallic particles.

After the characterization of the intermetallic particles, hydrogen evolution and electrochemical tests were performed to investigate the effect of Mn addition on the corrosion behavior of the Mg-3Al alloys. All of the corrosion tests were carried out in 3.5 wt.% NaCl solution with aeration at room temperature (around 25 °C), and the mechanical polished samples were adopted with 1 cm² exposed area. The hydrogen evolution tests were performed for 3 days by an apparatus [7]. For the electrochemical tests, the samples were sealed in epoxy and connected with copper lines. A platinum plate was used as the counter electrode, and the saturated calomel electrode was used as the reference electrode. Before the tests, the samples were held in NaCl solution for 10 min to reach a stable open circuit potential (OCP). The electrochemical impedance spectrum (EIS) tests were conducted by an electrochemical workstation (PARSTAT4000 type, Ametek Company, Berwyn, PA, USA). The EIS tests were carried out in the 100 kHz to 0.01 Hz frequency range at OCP with a 5 mV amplitude of perturbation. The EIS results were fitted using the ZsimpWin software (AMETEK, Berwyn, PA, USA). Potentiodynamic polarization tests were performed by an electrochemical workstation (CHI660E type, CH Instruments, Inc., Shanghai, China). The scan rate was about 0.5 mV/s with the range from −1.7 V to −1.3 V. Considering the abnormal anodic behavior brought from negative difference effect, the corrosion current density i_corr was obtained by Tafel extrapolation using the cathodic branch [58].

2.3. Thermodynamic Calculations

A new Mg-Al-Mn-Fe database was constructed based on the critical evaluation of the thermodynamic description on the binary and ternary systems by the present authors. There are six binary systems and four ternary systems in the Mg-Al-Mn-Fe system. For the Al-Mg system, the parameters were from Zhong et al. [59], who optimized the system based on their own first principles calculations and the available experimental thermodynamic and phase diagram data. The thermodynamic parameters of the Mg-Mn system was from Gröbner et al. [60], who investigated the system based on their own experimental data and available literature data. The thermodynamic description of the Mg-Fe system recorded by COST507 [61] was adopted in the present work because it was widely accepted in literature and not updated after that. The thermodynamic description of the Al-Mn system was taken from Du et al. [62], and the refinement by Zheng et al. [52] on the cubic Al₈Mn₅(HT) (Al₈Mn₅ in high temperature) phase was considered. As for the Al-Fe system, the most widely accepted evaluation is from Sundman et al. [63]. However, the description of the liquid phase was inconsistent with the experimental data on the experimental integral enthalpy of the mixing of liquid Al-Fe alloys. Therefore, the re-optimized parameters of liquid and the slightly revised parameters for the solids by Zheng et al [52] were adopted in the present work. Parameters of the Fe-Mn system from Huang et al. [64] with a minor adjustment by Djurovic et al. [65] were used in the present work. Since the work by Du et al. [62] could reproduce the experimental data in both Al-rich and Mg-rich corners well, and the liquidus projection and reaction scheme over wide temperature and composition ranges were also well established, the description for the Al-Mg-Mn system was taken the work from Du et al. [62]. The parameters of the Mg-Al-Fe ternary system were taken from the doctoral dissertation of Liu [66]. The Al-Fe-Mn ternary system was recently optimized by Zheng et al. [53]. In their work, Fe was firstly introduced into the sublattice model of Al₈Mn₅(LT) by considering the available experimental data [54], and they also considered the order-disorder transformation between Bcc_A2 and Bcc_B2 phases, thus their work was adopted for the description of the Al-Fe-Mn system. According to the work by Wang et al. [67], a direct extrapolation from the three binary systems can reproduce the experimental data in the Fe-Mg-Mn system well. Therefore, there is no ternary parameters for the Fe-Mg-Mn system. By considering the references above, we constructed a thermodynamic database for the Mg-Al-Mn-Fe quaternary system.

Using the obtained Mg-Al-Mn-Fe thermodynamic database, calculated phase diagrams for six binary systems and a representative isothermal section of four ternary systems are presented in Figure 1, Figure 2 and Figure 3. These diagrams show an agreement with those in literature, which support the validation of the database. Based on the database, the solidification paths of the Mg-3Al-xMn alloys were then presently performed by PANDAT^TM software (Version 2017, CompuTherm, Madison, WI, USA) using the Scheil solidification model.

3. Results and Discussion

3.1. Microstructure Observation

Back scattered electron (BSE) images of five electrolytic polished Mg-3Al-xMn alloys taken by EPMA are shown in Figure 4. The Mg-3Al alloy consists of (αMg) matrix and a minute quantity of white particles. In four Mn-added alloys, more white particles were observed at both (αMg) matrix and grain boundaries. The amount of these white particles increased gradually with the increase of Mn addition. The average grain size of the Mg-3Al-xMn alloys (x = 0, 0.12, 0.21, 0.36, 0.45) was measured by ImageJ software as 102 ± 35, 129 ± 48, 256 ± 75, 152 ± 37, and 205 ± 69 μm. Compared with the Mg-3Al alloy, the grain size of Mn-added alloys increased, indicating that the Mn addition had no grain refinement effect in the cast Mg-3Al alloys.

Quite a few studies have been applied to investigate the effect of Mn addition on the grain size of Mg-Al alloys. The grain refinement effect is related to the type of Al-Mn intermetallics in the alloys. Cao et al. [68] reported ε-AlMn could act as the nucleation site for (αMg) grains due to small lattice mismatch (~4% misfit) against (αMg), and Al₈Mn₅(LT) showed no grain refinement in their study due to the bigger lattice mismatch (~20% misfit). In addition, the grain refinement effect of Mn addition was observed in extruded or wrought Mg-Al alloys [18,19,50] and not observed in cast alloys [11,41,69]. This difference may relate to the thermal forming process. As is discussed later, the main Al-Mn intermetallic in present Mn-added alloys is Al₈Mn₅(LT), which cannot refine the grain. That is the reason why grain refinement effect was not observed in the present cast alloys.

To analyze the effect of Mn addition on the microstructure, it was crucial to identify the type and investigate the composition and morphologies of particles in the alloys. Figure 5 was determined by EPMA. The samples in this figure were mechanically polished. Figure 6; Figure 7 were obtained by SEM/EDX, and the samples were subjected to mechanical polishing, then deep-etched by nitric acid ethanol solution.

In the Mg-3Al alloy, irregular angular particles were observed in EPMA and SEM/EDX, which are shown in Figure 5a, Figure 6a, and Figure 7, respectively. Due to the limited amount, they were not successfully extracted and identified by XRD. For the phase composition of these particles, both EPMA and EDX maps indicated they were Mn negligible or Mn free, meaning they were Al-Fe phases. It should be mentioned that EPMA analysis could pick up Mg signals from surrounding (αMg) due to the small size of the particles, which was the reason why some Mg was detected. As shown in Figure 5a, the Al/Fe ratio of the particles determined by EPMA was about 3.25, which was close to the composition of Al₁₃Fe₄. According to the Mg-Fe phase diagram and isothermal section of Mg-Al-Fe shown in Figure 1c and Figure 3b, respectively, we can see that the solid solubility of Fe in Mg was very limited, indicating that the Fe atoms would precipitate as Al-Fe phases [22,70,71].

After adding 0.12 wt.% Mn into the Mg-3Al alloy, the intermetallics changed from Al-Fe phases to Al₈Mn₅(LT). XRD patterns of the particles extracted from the Mn-added alloys are shown in Figure 8. As we can see, Al₈Mn₅(LT) (space group R3m (160) ICDD PDF file 00-032-0021) and MgF₂ (space group P4₂/mnm (136) ICDD PDF 00-041-1443) were identified in the Mg-3Al-0.12Mn and Mg-3Al-0.21Mn alloys. The MgF₂ was likely formed from the reaction of the Mg melt with the protection of SF₆ gas during the sample fabrication [32,72,73]. To investigate the composition of Al-Mn(-Fe) particles, their Al/(Mn + Fe) ratios were measured by EPMA and are summarized in Figure 9b. For the Mg-3Al-0.12Mn and Mg-3Al-0.21Mn alloys, the Al/(Mn + Fe) ratios of these particles were measured to be from 1.2 to 1.9. The Al/Mn ratio of Al₈Mn₅(LT) was at the range of 0.9–1.6 based on the partial Al-Mn phase diagram shown in Figure 9a. Moreover, the measured Al/(Mn + Fe) of Al₈Mn₅(LT) was reported up to 1.97 [74] or 1.95 [31] in literature. In view of these results, these particles were identified as Al₈Mn₅(LT) in the present work, which was consistent with the XRD results.

As we can see from Figure 8, τ–Al_0.89Mn_1.11 was observed with further addition of Mn to 0.36 wt.%. Al₈Mn₅(LT), and τ–Al_0.89Mn_1.11 (space group tP2 (123) ICDD file 00-030-0028) was identified in the Mg-3Al-0.36Mn and Mg-3Al-0.45Mn alloys. The crystal structures of the crucial phases are summarized in

Table 2. Crystal structure of some crucial phases in the present study.

Phase	Other Name	Crystal System	Pearson Symbol	Space Group	Lattice Parameters		Reference
					a, b, c/Å	α, β, γ
Al8Mn5(LT)	γ2 D810-Al8Mn5	Hexagonal	$hR52$	$R3m$	12.54, 12.54, 15.74	α = β = 90°, γ = 120°	[75]
Al8Mn5(HT)	γ1 D82-Al8Mn5	Cubic	$cI52$	$I\overline{4}3m$	8.89, 8.89, 8.89	α = β = γ = 90°	[74,76]
τ–Al0.89Mn1.11	τ	Tetragonal	$tP2$	$P4/mmm$	2.77, 2.77, 3.54	α = β = γ = 90°	[77,78]
Al11Mn4(LT)	-	Triclinic	$ap15$	$P\overline{1}$	5.11, 8.87, 5.06	α = 89°, β = 101°, γ = 105°	[79]
B2-Al(Mn,Fe)	Bcc_B2, B2-AlMn	Cubic	$cP2$	$Pm\overline{3}m$	3.08, 3.08, 3.08	α = β = γ = 90°	[74,80]
(βMn)	-	Cubic	$cP20$	${\mathrm{P}4}_1$32	6.29, 6.29, 6.29	α = β = γ = 90°	[81]
MgF2	-	Tetragonal	$tP6$	${\mathrm{P}4}_2/mnm$	4.63, 4.63, 3.06	α = β = γ = 90°	[82]

. It should be noted that no peaks of MgF₂ were detected in the Mg-3Al-0.36Mn and Mg-3Al-0.45Mn alloys. τ–Al_0.89Mn_1.11 is a metastable phase and is not shown in the Al-Mn stable phase diagram (Figure 9a). Han and Liu [31] identified τ–Al_0.89Mn_1.11 via TEM and XRD, and the Al/(Mn+Fe) ratio was reported to be around 0.85, which was different from the range of both Al₈Mn₅(LT) and (βMn). In the present work, the measured Al/(Mn + Fe) ratio of some particles in the Mg-3Al-0.36Mn and Mg-3Al-0.45Mn alloys was about 0.85, as shown in Figure 9b. Thus, these particles were regarded as τ–Al_0.89Mn_1.11, as shown in Figure 5(d3,e2). Additionally, a lower Al/(Mn + Fe) ratio of Al₈Mn₅(LT) in the Mg-3Al-0.36Mn and Mg-3Al-0.45Mn alloys compared to that in the Mg-3Al-0.12Mn and Mg-3Al-0.21Mn alloys was obtained, which could be dated back to the solidification process. During the solidification, the Al concentration at the liquid/solid interface of the alloys with lower Mn content was relatively lower than that with higher Mn content, leading to the formation of Al₈Mn₅(LT) and τ–Al_0.89Mn_1.11 with a lower Al/(Mn + Fe) ratio [31]. Interestingly, some Al-Mn(-Fe) particles with relatively higher Fe content were obtained in the Mg-3Al-0.12Mn and Mg-3Al-0.45Mn alloys, as presented in Figure 5(b3,e3), respectively. Compared with the composition of B2-Al(Mn,Fe) in literature [74], these Fe-richer Al-Mn(-Fe) particles were likely to be B2-Al(Mn,Fe).

As for the morphologies, the particles in Figure 6b,c were indexed as Al₈Mn₅(LT), as it was the only Al-Mn(-Fe) intermetallic detected by XRD in Mg-3Al-0.12Mn and Mg-3Al-0.21Mn alloys. However, the particles in Figure 6d,e were difficult to distinguish from Al₈Mn₅(LT) and τ–Al_0.89Mn_1.11, because both phases were detected by XRD in Mg-3Al-0.36Mn and Mg-3Al-0.45Mn alloys.

Based on the above discussion, the morphologies and the Fe content of the particles in the Mg-3Al-xMn alloys are summarized in

Table 3. Summary of the phases, Fe content, and morphologies of the Al-containing intermetallic particles based on EPMA and SEM analysis.

Phase	Fe Range (at.%)	Morphologies	Alloys
Al8Mn5(LT)	0.2–4	Micro-plates Rods/ribbons Dice-like particles Polyhedral particles	Mg-3Al-xMn (x = 0.12, 0.21, 0.36, 0.45)
τ–Al0.89Mn1.11	0.5–5.9	Micro-plates Polyhedral particles	Mg-3Al-xMn (x = 0.36, 0.45)
B2-Al(Mn,Fe)	12.5–17.3	Micro-plates Polyhedral particles	Mg-3Al-xMn (x = 0.12, 0.45)
Al-Fe phase	15.3–17.3	Irregular blocky particles	Mg-3Al

. The Al₈Mn₅(LT) particles showed different morphologies in Mn-added alloys, including micro-plates (Figure 5(b2,c3,d1,e1)), rods/ribbons (Figure 5(b1,c1) and Figure 6(b1,c1), dice-like shape (Figure 6(c3)), and polyhedral shape (Figure 5(c2,d2), Figure 6(b2,b3,c2)). τ–Al_0.89Mn_1.11 with micro-plates (Figure 5(e2)) along with polyhedral shape (Figure 5(d3)) and B2-Al(Mn,Fe) with micro-plates (Figure 5(b3)) and polyhedral shape (Figure 5(e3)) were observed. Al-Fe phases were in irregular blocky shape (Figure 5a, Figure 6a, and Figure 7a). As shown in Figure 9c, the Fe content in Al₈Mn₅(LT) was measured to be at the range of 0.2–4 at.%, which was similar to the reported value in literature [31,34]. The Fe content at the range of 0.5–5.9 at.% was measured for τ–Al_0.89Mn_1.11, which was higher than the reported one by Han et al. [41]. In addition, it should be noted that the Fe content in B2-Al(Mn,Fe) and Al-Fe phases was significantly higher than that in Al₈Mn₅(LT) and τ–Al_0.89Mn_1.11.

The morphologies and the composition of the Al-containing intermetallic particles in the Mg-3Al-xMn alloys were discussed above. However, due to the limited amount, the percentages of these intermetallic phases could not be determined experimentally, and the sequences of solidification of these phases were still not clear. Such problems could be explained by the thermodynamic calculations.

3.2. Thermodynamic Calculation

As discussed before, the tolerance limit of Fe impurity is very important in Mg alloys. The Fe/Mn ratio of all alloys in the present work exceeded the tolerance limit of that in AM60 (0.021). Thus, it was worthwhile to trace the Fe impurity in the alloys. The calculations were performed based on the real composition of the Mg-3Al-xMn alloys. With the established database, Figure 10 shows the calculated solidification paths of the Mg-3Al-xMn alloys using PANDAT^TM software based on Scheil solidification. For the Mg-3Al alloy, B2-Al(Mn,Fe), (αMg), Al₂Fe, Al₅Fe₂ Al₁₃Fe₄, and Al₁₂Mg₁₇ phases were predicted to form in order. The Fe impurity was calculated to exist in the form of B2-Al(Mn,Fe), Al₂Fe, Al₅Fe₂, and Al₁₃Fe₄ phases at different stages of solidification. This calculation was in good agreement with the results from EPMA and SEM/EDX, which can be seen from Figure 5a, Figure 6a and Figure 7.

In Mn-added alloys, the sequences of solidification paths of the phases were generally similar to each other. The phases of B2-Al(Mn,Fe), (αMg), Al₈Mn₅(LT), and Al₁₁Mn₄(LT) were precipitated in sequence, which was significantly different from that of the Mg-3Al alloy. Moreover, in Mn-added alloys, impure Fe dissolved into B2-Al(Mn,Fe) and Al₈Mn₅(LT) phases by substituting the Mn lattice sites [55,56,74]. This was due to their affinity and similarity in terms of atomic radii [41,83]. From the thermodynamic point of view, B2-Al(Mn,Fe) could be stable in all five alloys because it was the first precipitated phase in all five alloys and could be encapsulated by the subsequent precipitated phases, such as Al₈Mn₅(LT). However, B2-Al(Mn,Fe) was only experimentally detected in the Mg-3Al-0.12Mn and Mg-3Al-0.45Mn alloys due to the limited amount. The deduction was supported by the experiments by Zeng et al. [74] and present thermodynamic calculations. Two phase particles with a core-shell structure in AZ91D magnesium alloy with the Fe-richer B2-Al(Mn,Fe) core enveloped by low-Fe Al₈Mn₅(LT) shell were observed by means of a focused ion beam (FIB) from Zeng et al. [74]. Considering the predicted solidification sequences and the similar morphology of Al₈Mn₅(LT) from the present work and Zeng et al. [74], a similar enveloping phenomenon likely existed in this study.

The total mole percentages of B2-Al(Mn,Fe) and Al₈Mn₅(LT) of these five alloys after solidification were calculated and are displayed in Figure 11a. It can be seen that Al₈Mn₅(LT) was not precipitated in the Mg-3Al alloy, while the mole percentage of B2-Al(Mn,Fe) held steady with 0–0.36 wt.% Mn addition but increased when the content of Mn addition was 0.45 wt.%. The reason for this was that the Fe impurity in the Mg-3Al-0.45Mn alloy was higher than that of other alloys, as shown in Table 1. For Al₈Mn₅(LT), the percentage increased steadily with the increasing of Mn addition, which was consistent with the statistic number of particles per mm² (Figure 11b) based on the low magnification EPMA images. Furthermore, the mole percentage of B2-Al(Mn,Fe) was far less than that of Al₈Mn₅(LT). This was one of the reasons why B2-Al(Mn,Fe) was not detected by EPMA in the Mg-3Al, Mg-3Al-0.21Mn, and Mg-3Al-0.36Mn alloys. More importantly, although the amount of Fe-richer B2-Al(Mn,Fe) seemed negligible due to the detrimental effect, it played an important role in the corrosion of Mg-Al-based alloys [28].

3.3. Electrochemical Tests and Corrosion Analysis

Potentiodynamic polarization curves are shown in Figure 12, which were tested in 3.5 wt.% NaCl solution at room temperature. The cathodic branches of the polarization curves of Mn-added alloys were lower than those of the Mg-3Al alloy, which indicated that there was a reduction of the cathodic hydrogen reaction. Obvious passivation behavior was observed in the anodic polarization curves of the Mg-3Al-0.36Mn and Mg-3Al-0.45Mn alloys, however, curves of others alloys showed active dissolution characteristics. The breakdown potential E_b (marked by the black arrow) of passive film was also obtained [58,84]. Critical parameters from potentiodynamic polarization curves are summarized in

Table 4. Critical parameters determined from polarization curves for the Mg-3Al-xMn alloys.

Alloys	icorr/uA·cm−2	Ecorr/VSCE	Eb/VSCE	(Eb − Ecorr)/mV
Mg-3Al	410.1	−1.560	-	-
Mg-3Al-0.12Mn	112.5	−1.545	-	-
Mg-3Al-0.21Mn	51.3	−1.546	-	-
Mg-3Al-0.36Mn	18.6	−1.485	−1.471	14
Mg-3Al-0.45Mn	15.8	−1.512	−1.417	95

. With the increase of Mn addition, the current density i_corr decreased gradually, and the value of the Mg-Al alloy was 26 times larger than that of the Mg-3Al-0.45Mn alloy. The larger value of (E_b-E_corr) illustrated more stable passivation film. The value of Mg-3Al-0.45Mn was 94 mV, which was markedly larger than that of Mg-3Al-0.36Mn, indicating that the passive film was more stable, and there was less tendency for localized corrosion to appear [84]. In summary, the corrosion resistance was enhanced with the increase of Mn addition, and the Mg-3Al-0.45Mn alloy showed the best corrosion resistance among the five alloys.

The EIS results presented in Figure 13 show all alloys consisted of two capacitance loops and one inductive loop, indicating that the corrosion mechanisms and processes were similar among these alloys. The first capacitance loop at high frequency represented the double electric layer at the metal/solution interfaces, while the second one at medium frequency was the capacity of the corrosion products. The inductive loop at low frequency confirmed pitting corrosion behavior [85,86]. The aperture of impedance increased with the enhancement of Mn content, which indicated that the corrosion resistance was improved [18]. The EIS spectrums were fitted by an equivalent circuit [18,85], as shown in Figure 13b and

Table 5. Fitting results of the EIS data for the Mg-3Al-xMn alloys.

Specimen	Rs /Ω·cm−2	Cc /μF·cm−2	Rc /Ω·cm−2	CPE		Rct /Ω·cm−2	L /H·cm−2	Rdiff /Ω·cm−2
				C /μF·cm−2	n (0–1)
Mg-3Al	4.81	6.62	12.75	3.05	1	21.34	14.46	78.9
Mg-3Al-0.12Mn	2.91	8.38	10.26	11.12	0.9123	37.03	26.42	86.4
Mg-3Al-0.21Mn	1.03	7.34	29.14	51.34	0.3518	53.95	58.68	255.4
Mg-3Al-0.36Mn	2.99	5.99	22.01	6.23	0.8646	95.21	49.58	279.5
Mg-3Al-0.45Mn	4.73	7.08	20.76	49.23	0.5554	106.54	58.09	341.9

. R_s was the solution resistance, and R_c and C_c represented the resistance and the capacitance of the corrosion product layer, respectively. CPE was the constant phase element for double layer capacitance, and its impedance was defined as $Z_{\mathrm{CPE}}={\left[C{\left(j\omega \right)}^n\right]}^{-1}$, where C was capacitance, j was the current, and ω was the frequency [18]. R_diff and L denoted the inductive resistance and the inductance of the partial protective film. R_ct was the charge transfer resistance describing the resistance of interface between corrosion product layer and alloy, which was the last barrier for preventing aggressive Cl⁻ in the solution from reaching the metal surface [85,87]. A larger R_ct meant a better corrosion resistance. The values of R_ct became larger with the increasing of Mn addition, meaning the Mg matrix corrosion could be prevented by the Mn addition in the alloys. This was also supported by the polarization curves discussed above.

Figure 14 shows the effect of Mn on hydrogen evolution volume of the Mg-3Al-xMn alloys in 3.5 wt.% NaCl solution. In three days immersion, the hydrogen volume of the Mg-3Al alloy was about three to eleven times that of the Mn-added alloys, demonstrating that Mn addition improved corrosion resistance by decreasing the cathodic hydrogen reaction. The hydrogen volume of the alloys decreased gradually with the increase of Mn, which was in agreement with the electrochemical tests. As discussed above, due to the limited amount of Mn, the Fe impurity precipitated as the Al-Fe phases in the Mg-3Al alloy. The micro-galvanic corrosion between Al-Fe phases and the (αMg) matrix was extremely serious. After adding some Mn into the alloys, B2-Al(Mn,Fe), Al₈Mn₅(LT), and τ–Al_0.89Mn_1.11 precipitated with Fe. Although there was still micro-galvanic corrosion between these Al-Mn(-Fe) particles and the (αMg) matrix [32,88], this detrimental effect was suppressed significantly. There were two reasons for the improvement. First, the potential difference of Al-Mn(-Fe) particles was significantly less than Al-Fe particles with respect to the magnesium matrix [11,22,37,89]. Second, as shown in the above thermodynamic calculations, the first precipitated Fe-rich B2-Al(Mn,Fe) could be encapsulated by Al₈Mn₅(LT), which prevented the contact of the Fe-rich phase and the (αMg) matrix [74]. Additionally, the increasing amount of Al₈Mn₅(LT) with the increment of Mn content alleviated the harmful effect gradually. Thus, Mn addition improved the corrosion behavior of the Mg-Al alloys by changing the type and the amount of the intermetallic particles.

4. Conclusions

The effects of Mn addition on microstructure and corrosion behavior of the Mg-3Al-xMn (x = 0.12, 0.21, 0.36, 0.45) alloys were investigated through the combination of experiments and thermodynamic calculations. The main conclusions drawn are as follows:

(1) Mn addition did not show the grain refinement on the cast Mg-3Al-xMn alloys. A small amount of the Al-Fe particles was observed in the Mg-3Al alloy. Al₈Mn₅(LT) existed as the main intermetallic compound in all Mn-added alloys. Only in the Mg-3Al-0.36Mn and the Mg-3Al-0.45Mn alloys, a lower Al/(Mn + Fe) ratio τ–Al_0.89Mn_1.11 phase was observed.

(2) Based on the Scheil solidification calculation, Fe-richer B2-Al(Mn,Fe) was predicted to form as the first precipitated phase in all alloys. Due to the limited amount, B2-Al(Mn,Fe) was only experimentally observed in the Mg-3Al-0.12Mn and Mg-3Al-0.45Mn alloys. Al₈Mn₅(LT) would envelop B2-Al(Mn,Fe) in Mn-added alloys.

(3) Pitting corrosion characteristic was observed from the polarization and EIS curves in all alloys. Passivation stages were only detected in the polarization curves of the Mg-3Al-0.36Mn and the Mg-3Al-0.45Mn alloys. Both the electrochemical and the hydrogen evolution tests indicated that Mn addition gradually improved the corrosion resistance of the Mg-Al alloys in NaCl solution. The high cathodic Al-Fe intermetallic compounds could be changed into Al₈Mn₅(LT) by adding minor quantities of Mn (0.12 wt.%). Increasing Mn addition would lead to more Al₈Mn₅(LT), which could encapsulate the Fe-richer B2-Al(Mn,Fe) phase and thus alleviate micro-galvanic corrosion in the alloys.