2020 年 61 巻 9 号 p. 1881-1888
The corrosion behavior of diecast Mg–Al–Mn–Ca–Si (AMXS6020) magnesium alloy with superior flame-retardance and heat-resistance was systematically investigated and compared to diecast AZ91D magnesium alloy and ADC12 aluminum alloy by salt spray test, cyclic corrosion test, electrochemical experiment, scanning electron microscopy and energy dispersive X-ray spectroscopy. The results showed that AMXS6020 alloy had a higher corrosion resistance than AZ91D and ADC12 alloys. The superior corrosion resistance of AMXS6020 alloy was attributed to netlike eutectic structure consisting of (Al, Mg)2Ca and α-Mg, which could hinder the corrosion growth. The barrier effect of netlike (Al, Mg)2Ca phase structure on corrosion growth was also confirmed by potentiodynamic polarization.
Fig. 5 Pictures of as-diecast specimens of Mg and Al alloys before and after SST for 96 hours.
Since magnesium is the lightest metal among practical metals, magnesium and its alloys are expected to be used in handheld tools, mechanical parts and automobiles where weight saving is important. However poor corrosion resistance and flammability of magnesium alloys are obstacles to their wide applications.1,2) Since Pilling-Bedworth ratio3) of magnesium is less than 1, oxide products could not cover the surface completely. Therefore the surface of magnesium alloys would oxidized easily, and they always ignite in melting or casting process. Fortunately it has been found that addition of calcium to magnesium alloy can suppress the ignition of magnesium alloys.4,5) The formation of ignition-protective layer on the surface of Mg alloys has been reported. AMX602 is flame-retardant magnesium alloy added about 2 mass% of calcium to AM60 magnesium alloy.
When Mg alloys are applied to automotive parts such as power-train that are usually exposed to elevated temperature environment, good heat resistance (i.e. creep property) is required. However, the heat resistance of commercially available Mg alloys is generally inferior to that of Al alloys. For example, AZ91D is the most popular diecast Mg alloy having comparatively good castability and mechanical strength, but its heat resistance is so poor that it cannot be used for automotive power-train. To improve the heat resistance of Mg alloys, Y, Si and Ca elements are added into Mg alloys.6–8) A flame-retardant and heat-resistant Mg–Al–Mn–Ca–Si alloy has been developed and its creep strain is reported to be less than 0.15% at the stress of 50 MPa and temperature of 448 K after 100 hours.9)
The poor corrosion resistance of magnesium alloys is attributed to the less-protective hydroxide film formed on magnesium surface and the internal galvanic corrosion caused by intermetallic phases or impurities.10) The influence of Mg17Al12 intermetallic compound (β-phase) in AZ91D on the corrosion behavior is complex and dependent on their type, size and morphology.11–15) Many researches show that β-phase can act either as a barrier to inhibit the corrosion or as a galvanic cathode to accelerate the corrosion of magnesium alloys.12,15) The β-phase is expected as a barrier when there is a small grain size and relatively large β-phase fraction, and more importantly the β-phase is in the form of a continuous network along Mg grain boundaries. In contrast, microgalvanic corrosion is readily accelerated as the β-phase is agglomerated and separately distributed in the Mg matrix of coarse grains.12) Other intermetallics than β-phase are found to be harmful by promoting the microgalvanic corrosion.16,17) While almost all the intermetallic phases present in the commercial magnesium alloys exhibit nobler potentials than magnesium matrix, their potential difference with magnesium matrix is various for different phases.18,19) Regarding Ca-containing Mg alloys, some studies have been performed to investigate the corrosion behavior of gravity cast Mg–Al–Zn–Ca20,21) and Mg–Al–Mn–Ca systems,22,23) and revealed that continuously formed Al–Ca intermetallics can interrupt the propagation of corrosion.21) The corrosion behavior of Si-containing Mg alloys like Mg–Al–Si systems has also been studied.24,25) It has reported that the corrosion resistance of Mg–Al–Si alloys is deteriorated due to the coarsened β and Mg2Si phases.25)
The corrosion resistance of Mg–Al–Mn–Ca–Si alloy (AMXS6020) alloy is not well known because that has both systems of Al–Ca that would be beneficial for corrosion resistance and Mg–Si systems that might be harmful to corrosion resistance. In the present study, the corrosion resistance of AMXS6020 diecast alloy was investigated and compared to that of AZ91D and ADC12 diecast alloys, with the focus on the influence of intermetallic phases. Diecast ADC12 is the most widely employed Al alloy in automobile industry, so it can be a benchmark for evaluating the corrosion resistance of Mg alloys.
AMXS6020, AZ91D and ADC12 diecast alloy plates with dimensions of 70 × 130 × 2 mm were used in the present work. The chemical compositions of these Mg and Al alloys are given in Table 1.
The microstructures of AMXS6020, AZ91D and ADC 12 diecast alloys were examined on cross-sections of specimens using an optical microscope (Nikon GX51) and scanning electron microscope (SEM, JEOL JSM-7000F). Specimens for microstructure observation were finally polished using 0.25 µm diamond suspension, and then etched using dilute picric acid or dilute hydrofluoric acid for Mg or Al alloys, respectively. Secondary electron image (SEI) mode was used in SEM observation. Compositional contrast (COMPO) images of microstructures were obtained with backscattered electron imaging equipped to the SEM. The chemical compositions were analyzed using an energy dispersive X-ray spectrometer (EDS, JEOL JED-2300) equipped to the SEM. Elemental maps were obtained with the EDS system. Acceleration voltage was 15 kV. Intermetallic phases in the Mg and Al alloys were identified by X-ray diffraction (XRD, Rigaku Ultima IV) using Cu Kα radiation with a step size 0.02°.
2.3 Corrosion testsCorrosion tests were performed using cyclic corrosion test (CCT) and salt spray test (SST) according to International Organization for Standards ISO 11997-126) and ISO 9227,27) respectively. CCT was performed using “Cycle A” which a cycle consists of salt fog at 35°C for 2 hours, drying at 60°C/RH 20∼30% for 4 hours and humidity at 50°C/RH ≥95% for 2 hours. The test period was 96 hours (12 cycles). SST was performed according to “Neutral salt spray test” at 35°C. The test time was 96 hours. 5 mass% NaCl aqueous solution was used both CCT and SST. The specimens with superficial layer removed were subjected to CCT, and the specimens with diecast surface were subjected to CCT and SST. For the specimens with superficial layer removed, the surface of specimens was finally ground with #1200 abrasive papers. The exposure area for the CCT and SST was 50 × 100 mm, and the remaining region was covered with masking tape.
After the corrosion tests, the masking tape was peeled off, and the corrosion products formed on the surface of specimens were removed as follows: AMXS6020 and AZ91D specimens were boiled in 10% CrO3 aqueous solution for 1 minute, while ADC12 specimens were boiled in an aqueous solution containing 2% of CrO3 and 5% of H3PO4 for 10 minutes. The corrosion rate was calculated by measuring the weight loss of specimens before and after corrosion tests. Average value of 3 specimens was used to determine the corrosion rate of each alloy. In order to avoid the error induced by the above corrosion product treatment, specimens without corrosion tests but subjected to the above corrosion product treatment were used as control.
In addition, the surface and cross-section of some specimens after corrosion tests were observed with SEM and the elemental maps were obtained in order to clarify corrosion behavior and mechanism.
2.4 Electrochemical measurementsRest potential measurement was carried out using Keithley 2000 digital multi-meter in 5 mass% NaCl aqueous solution without deaeration. The specimens for the test were finally ground with #1200 abrasive papers to remove the traces of die spray and superficial layers. The exposure area was 1 cm2.
Potentiodynamic polarization test was performed using Solartron 1280Z electrochemical measurement system. The specimens and electrolyte were the same as those for rest potential measurement. The anodic polarization sweeps were started at 0.2 V negative to their rest potential after 1 or 72 h immersion at the scan rate of 0.5 mV s−1. The reference electrode was Ag/AgCl using saturated KCl as internal solution. Platinum electrodes were used as counter electrodes. Polarization resistance was calculated from the relation of current – potential in a section of rest potential ±10 mV. All measurements were performed at room temperature in open air.
The optical microstructures of AMXS6020, AZ91D and ADC12 diecast alloys are presented in Fig. 1. All of the three alloys exhibited network-like structures, but the morphology and/or elaborateness of the network-like structure was different. AMXS6020 magnesium alloy and ADC12 aluminum alloy revealed continuous network-like structure, but the network-like structure in AZ91D alloy was discontinuous. This could be related with the formation process of the network-like microstructure. The network-like structure in AMXS6020 and ADC12 alloys was formed via eutectic reaction during solidification, while that in the AZ91D alloy was formed through precipitation after solidification. This will be described in detail later. In addition, the microstructure seemed to be finer in AMXS6020 alloy than in ADC12 alloy.
Microstructures of diecast Mg and Al alloys observed with optical microscope.
SEM images and elemental maps of AMXS6020, AZ91D and ADC12 diecast alloys are shown in Fig. 2. For AMXS6020 and AZ91D magnesium alloys, Mg, Al, Si and Mn maps are displayed. Ca map is also given for AMXS6020 alloy. Al, Si, Fe and Cu maps are displayed for ADC12 aluminum alloy. The difference in the microstructure among the three alloys could be seen more clearly from Fig. 2.
SEM images and elemental maps of Mg and Al diecast alloys.
In AMXS6020 alloy, the network was compact and continuous, in which Al and Ca were more contained than in α-Mg matrix as shown in Fig. 2. The network is composed of Al2Ca and (Al,Mg)2Ca intermetallic phases.28) XRD analysis indicated that the microstructure consisted mainly of the primary α-Mg matrix, while the intensity of peaks attributed to Al2Ca and (Al,Mg)2Ca was too weak to discern. The ratio of Al to Mg in the (Al,Mg)2Ca intermetallic phase was estimated 1.34:0.66. It was thought that the intensity of peaks attributed to (Al,Mg)2Ca was very weak because the amount of this phase was small. The (Al,Mg)2Ca was formed principally along the boundaries of α-Mg dendrites via eutectic reaction of Al and Ca elements during solidification, where Mg element was also involved.23) The wall thickness of netlike (Al,Mg)2Ca phase was about 0.2 µm. Mg–Ca–Si intermetallic phase has been reported elsewhere,28) but it was not identified in this work. This is probably because the amount of Mg–Ca–Si phase is so small that it cannot be detected by conventional XRD technique. To identify the Mg–Ca–Si phase, transmission electron microscopy and electron diffraction are necessary.28)
For AZ91D alloy, the microstructure was composed mainly of the primary α-Mg matrix and Mg17Al12 intermetallic phase (β-phase) (Fig. 3). The β-phase was precipitated along the α-Mg boundaries after solidification, forming network-like structure. However, the network-like structure of β-phase was not continuous (Fig. 2).
XRD patterns of (a) AMXS6020, (b) AZ91D and (c) ADC12.
In the XRD spectra of ADC12 alloy, peaks attributed to Al, Si and CuAl2 phases were detected (Fig. 3), suggesting that the microstructure of ADC12 alloy was mainly primary Al phase and Al–Si eutectic structure.29) That is to say, the Al and Si were present separately, and primary Al phase in ADC12 alloy seemed to connect each other through the network-like structure. This phenomenon was different from that observed in AMXS6020 alloy, where Al and Ca reacted to form network-like (Al,Mg)2Ca phase. It should be noted that Al–Fe–Mn–Si intermetallic phase were also included in ADC12 alloy, but the amount of these intermetallic phases was small.
Table 2 shows average chemical compositions of 9 and 8 analyzed areas in α-Mg matrix of AMXS6020 and AZ91D magnesium alloys, respectively. Some of analyzed area are shown in Fig. 2 by red marking. Al concentration in α-Mg matrix was higher in AZ91D alloy than in AMX6020 alloy. The Al concentration in α-Mg matrix has an influence on the corrosion resistance of Mg alloys. This will be described later.
CCT and SST are often used to assess the corrosion resistance in practical application to automobile parts as simple acceleration test.
In order to eliminate the influence of superficial layer on the corrosion behavior of diecast alloys, the specimens with surface ground were employed in CCT. The results of three alloys are given in Fig. 4. The corrosion rate was in the order of AMXS6020 < ADC 12 < AZ91D. AMXS6020 alloy exhibited the lowest corrosion rate among the three alloys.
Corrosion rate of Mg and Al alloys in CCT for 96 hours.
In the present study, the corrosion rate of AZ91D alloy was higher than that of ADC12 alloy. It is well known that corrosion resistance of AZ91D alloy is affect by impurities such as Cu, Ni and Fe,30) so that the upper limits of Cu, Ni and Fe in AZ91D alloy are specified by International Organization for Standardization ISO 16220 and Japanese Industrial Standards JIS H2222. However, corrosion rate of AZ91D alloy could vary dramatically even though the impurity concentrations were in the range specified by the standards, depending on the microstructure of AZ91D alloy. The corrosion rate of AZ91D alloy evaluated by SST was reported to vary from 0.13 to 1.2 mg cm−2 day−1.14,30) This is to say, the corrosion rate of AZ91D alloy might be lower than that of ADC12 alloy in some cases.30)
To assess and compare the corrosion resistance of as-diecast AMXS6020 alloy with that of AZ91D and ADC12 alloys in practical use, SST and CCT were carried out using the specimens with diecast surface remained. The surface appearance change of as-diecast specimens before and after SST for 96 hours is displayed in Fig. 5. The photographs of specimen surface after SST were taken before the corrosion products on specimen surface were removed. Before SST, all specimens exhibited a similar appearance of as-cast surface, i.e. dull metallic surface. After 96 h of SST, the surface appearance of specimens was obviously different, depending on the type of alloy. Although the color of specimens was changed, the amounts of corrosion product calculated from weight loss of AMXS6020 Mg alloy was the fewest among the three alloys. AZ91D and ADC12 alloys showed a similar extent of corrosion, but the color of corrosion products was different.
Pictures of as-diecast specimens of Mg and Al alloys before and after SST for 96 hours.
The SST and CCT results of as-diecast specimens of the three alloys are given in Fig. 6. The order of corrosion rate was AMXS6020 < ADC 12 < AZ91D, irrespective of corrosion test type and surface condition of alloys. AMXS6020 alloy with diecast surface also exhibited the lowest corrosion rate among the three alloys as in the case the superficial layer was removed.
Corrosion rate of as-diecast specimens of Mg and Al alloys in SST and CCT for 96 hours.
It could be seen from Fig. 6 that the corrosion rate of AZ91D alloy was larger in SST than in CCT, while the corrosion rate of AMXS6020 and ADC12 alloys was not remarkably affected by corrosion test type. The corrosion rate of Mg alloys is usually proportional to the period in contact with water. CCT and SST are intermittent and continuous spray, respectively. Therefore, corrosion of Mg alloy proceeded more quickly in SST with continuous spraying than in CCT with intermittently spraying, resulting in a higher corrosion rate of AZ91D alloy in SST.
The corrosion rate of AMXS6020 Mg alloy was remarkably lower as compared to AZ91D and ADC12 alloys, and not obviously influenced by corrosion test type. From the point of view of aluminum concentration of α–Mg phase, AMSX6020 should be inferior to AZ91D at corrosion resistance, but the result was in contrary. This result is believed to be related to the network-like microstructure in AMXS6020 alloy. This will be described later.
3.3 Corrosion morphologyElemental maps on specimen surface of the three alloys after SST for 24 hours are presented in Fig. 7. Oxygen map was thought to correspond to corrosion products formed during SST. Corrosion products has crevices due to Pilling-Bedworth ratio of magnesium being less than 1. Arrows in Fig. 7 indicate corrosion products. The corrosion products on the surface of AZ91D and ADC12 alloys were observed in agglomeration. For AMXS6020 alloy, corrosion occurred inside network and corrosion progress was almost stopped by the network during 24 h SST, as seen from Fig. 7. This is because network-like structure was continuous and compact in AMXS6020 alloy. For AZ91D alloy, the network was not continuous so that adjacent corrosion products likely jointed together to form massive corrosion products. The network in ADC12 alloy did not function as a strong barrier against corrosion progress, because the netlike structure consisted of Al and Si phases instead of intermetallic phase, i.e. the principal phase in the matrix and netlike structure was the same. Consequently, corrosion of ADC12 alloy advanced easily during SST.
Elemental maps of surface of Mg and Al diecast alloys after SST for 24 hours.
Figure 8 presents elemental maps of cross section of the three alloys after CCT for 24 hours. The barrier effect of networks in AMXS6020 and AZ91D alloys could be observed more obviously. Because the network was incomplete in AZ91D, corrosion could spread through the incomplete network into adjacent grains. For ADC12 alloy, corrosion products covered the surface of specimen, and barrier effect of netlike structure was not observed.
Elemental maps of cross section of Mg and Al diecast alloys after CCT for 24 hours.
To investigate the electrochemical response of AMXS6020 alloy, the rest potential measurement and potentiodynamic polarization were performed, and the results were compared with those of AZ91D and ADC12 alloys. The variation of rest potential of the three alloys with immersion time in 5 mass% NaCl aqueous solution are shown in Fig. 9. The potentiodynamic polarization curves of three alloys after immersed in 5 mass% NaCl solution for 1 and 72 hrs are shown in Fig. 10.
Rest potential of Mg and Al diecast alloys during immersion.
Potentiodynamic polarization curves of (a) AMXS6020, (b) AZ91D and (c) ADC12.
AMXS6020 and AZ91D magnesium alloys showed c.a. −1.9 V vs. Ag/AgCl (the same reference hereinafter) of rest potential at the beginning of immersion. The rest potential of AZ91D alloy shifted to be c.a. −1.50 V after 1.8 ks immersion, and became comparatively stable at this value. In contrast, it took about 12 ks for the rest potential of AMXS6020 alloy to shift to be c.a. −1.50 V, and the rest potential of AMXS6020 alloy became comparatively stable at c.a. −1.53 V. The rest potential of both AMXS6020 and AZ91D alloys shifted by 0.2–0.3 V and turned back to the stable values in a short period of time during immersion. For ADC12 alloy, the rest potential became comparatively stable at c.a. −0.69 V immediately after start of immersion.
The rest potential is a mixed potential determined by cathodic and anodic reactions. In this case, the cathodic reaction of Mg alloys is mainly the reduction of water in neutral solution, and anodic reaction is the dissolution of Mg alloy. The mechanism about the rising of rest potential at the early stage of immersion is thought as following. Hydroxide film might form on α-Mg phase during immersion.22) α-Mg phase acts as anodic site, so that covering with hydroxide film of anodic site causes anodic reaction decreased. As a result, the rest potential shifted toward noble direction. The vibration of the rest potential of AMXS6020 and AZ91D alloys in a short period of time during immersion might be related to the dissolution of network-like-intermetallic phases, but the mechanism has not been understood yet.
As shown in Table 2, Al concentration in α-Mg matrix was higher in AZ91D alloy than in AMXS6020 alloy. It has been reported that the corrosion of Mg alloys likely occurs in α-Mg matrix, and corrosion resistance of Mg alloys increases with Al concentration in α-Mg matrix.23,31) Therefore, it could be thought that higher Al concentration in α-Mg matrix of AZ91D alloy suppresses its anodic reaction in aqueous solution, so the rest potential of AZ91D alloy was higher than that of AMXS6020 alloy. It could also be seen from Fig. 9 that the stabilized rest potential of ADC12 aluminum alloy was much higher than that of AZ91D and AMXS6020 magnesium alloys. The high rest potential of ADC12 alloy was believed to be attributed to its much higher Al content in ADC12 alloy.
In potentiodynamic polarization curves as displayed in Fig. 10, the corrosion current density of AMXS6020, AZ91 and ADC12 alloys after 1 h immersion, which was measured using Tafel extrapolation of the cathodic polarization curves to corrosion potential, was 1.46 × 10−5, 4.60 × 10−6 and 5.96 × 10−8 A/cm2, respectively. These results indicated that the corrosion rate of the three alloys is in order of AMXS6020 > AZ91 > ADC12 at the initial stage of corrosion, although the corrosion current density cannot be estimated reliably by Tafel extrapolation.32)
When the immersion time was prolonged from 1 to 72 hrs, the corrosion current density of AMXS6020 alloy decreased remarkably as seen in Fig. 10(a), while the corrosion current density of AZ91D and ADC12 alloys increased as seen in Fig. 10(b), (c). For AMXS6020 alloy, anodic current was strongly suppressed after 72 h immersion and this seemed to be effective in reducing the corrosion current density. A rise in the anodic current that corresponding to deterioration of hydroxide layer was observed, this suggests that the hydroxide layer exists on the surface at rest potential. The potential in which anodic current rises shifted toward noble direction after 72 h immersion. This implies the growth of hydroxide layer during the immersion. In addition, the network-like (Al,Mg)2Ca phase might inhibit corrosion progression and depressed α-Mg solution. For AZ91 and ADC12 alloys, the increase of corrosion current density is believed to be caused by the increment of exposed area of α-Mg after immersion for 72 hrs. It should be noted the corrosion current density might be effected by enlargement of surface area owing to dissolution after 72 h immersion.
3.5 Barrier effect of netlike structure on corrosion processFrom above results and discussion, the corrosion process of the three alloys used in the present study could be described by the schematic diagram shown in Fig. 11. At the early stage, corrosion occurred in α-Mg matrix of AMXS6020 and AZ91D alloys and in both the matrix and eutectic structure of ADC12 alloy. The netlike structures in AMXS6020 and AZ91D alloys were composed of intermetallic compounds (IMCs) which had comparatively high corrosion resistance, whereas the eutectic structure of ADC12 was Al and Si phases whose corrosion resistance was at the same level as Al matrix. Because the Al concentration was lower in the α-Mg matrix of AMXS6020 alloy than AZ91D alloy, the corrosion resistance of the three alloys should be in order of AMXS6020 < AZ91D < ADC12 at the early stage of corrosion. With the advance of corrosion, the α-Mg matrix on the surface of AMXS6020 and AZ91D alloy specimens was dissolved and the corrosion progress was inhibited by the boundaries of netlike structures, but eutectic network-like structure in ADC12 alloy had almost no influence on the corrosion progress. At the later stage, corrosion progress of AMXS6020 alloy was remarkably delayed by the continuous boundaries of network-like structures; however, corrosion advanced through the gap where the network-like structures were not continuous for AZ91D alloy. As a result, AMXS6020 alloy had much higher corrosion resistance than AZ91D and ADC12 alloys.
Schematic diagram of the corrosion process of diecast Mg and Al alloys.
The corrosion resistance of diecast AMXS6020 magnesium alloy with superior flame-retardance and heat-resistance was investigated using SST, CCT, electrochemical measurement and compared to that of diecast AZ91D magnesium alloy and ADC12 aluminum alloy. The results showed that AMXS6020 alloy had a much higher corrosion resistance than AZ91D and ADC12 alloys, although the corrosion resistance was in order of ADC12 > AZ91D > AMXS6020 at the initial stage of corrosion. The superior corrosion resistance of AMXS6020 alloy was found to be attributed to eutectic intermetallic compound of (Al,Mg)2Ca phase which forms continuous network-like structure. Network-like structures were also observed in AZ91D and ADC12 alloys, but the network in AZ91D alloy was not continuous and principal phase in the network of ADC12 alloy was the same as in matrix. The network-like structure of (Al,Mg)2Ca phase in AMXS6020 alloy remarkably inhibited corrosion progress, so that the corrosion resistance was greatly improved.
