MATERIALS TRANSACTIONS
Online ISSN : 1347-5320
Print ISSN : 1345-9678
ISSN-L : 1345-9678
Mechanism of Intergranular Corrosion of Brazed Al–Mn–Cu Alloys with Various Si Content
Michihide YoshinoShohei IwaoMasakazu EdoHajime Chiba
Author information
JOURNALS FREE ACCESS FULL-TEXT HTML

2017 Volume 58 Issue 5 Pages 768-775

Details
Abstract

This study investigated the effect of post-brazing cooling rate and Si addition on the intergranular corrosion (IGC) susceptibility of brazed Al–Mn–Cu alloys by electrochemical analysis and microstructure observation. Water-quenched samples after brazing exhibited no IGC susceptibility, whereas slowly-cooled samples were prone to IGC. The results suggest that IGC is caused by precipitation during cooling. In addition, it was observed that IGC susceptibility depended on the Si content. An alloy sample with a low Si-additive content exhibited high IGC susceptibility because Mn/Cu-depleted zone was formed near the grain boundaries as a result of the preferential precipitation of Al6(Mn,Fe) and CuAl2 on the grain boundaries. In contrast, moderate Si addition inhibited IGC because the decrease of the Mn content in the grain interiors due to enhanced precipitation of Al15(Mn,Fe)3Si2 in the grain. Additionally, Cu-depleted zone also disappeared because preferential precipitation of CuAl2 on the grain boundaries was prevented. The excess-Si alloy exhibited high IGC susceptibility because Si-depleted zone formed around the grain boundaries as a result of the preferential precipitation of coarse Si particles on the grain boundaries although the Mn/Cu-depleted zones were not formed.

1. Introduction

Al–Mn system alloys, typified by A3003, have been used for automotive heat exchangers because of their moderate strength, good formability, and superior corrosion resistance. Recently, thinner and stronger materials have been required to satisfy weight-saving and high-performance requirements for heat exchangers1,2). As a result, modified A3003 alloys with high Mn, Cu, and Si contents have been used in this application. Tube stock is produced from thin material by roll forming and constitutes the main component of heat exchangers. The stock must resist penetration and leakage of the content inside it due to corrosion. Intergranular corrosion (IGC) is the worst corrosion morphology because it enables easy penetration due to local attack at grain boundaries. Therefore, IGC resistance is an essentially technical requirement for the material used in heat exchangers.

In general, non-heat-treatable Al–Mn alloy exhibits good corrosion resistance and no IGC susceptibility. However, IGC occurs when the Al–Mn alloy is brazed at a low cooling rate3). In addition, IGC susceptibility tends to increase with increasing the content of Mn and Cu3,4). In contrast, IGC is reportedly inhibited by Si5,6). Recently, Si has become a popular additive because it tends to improve the strength of Al–Mn based alloys. Therefore, understanding the effect of Si addition is important from a practical viewpoint. Metallographic investigations of the IGC mechanism on this alloy system have been insufficient. Although studies on Al–Mn alloy7) and Al–Mn–Cu alloy8) have been reported, the literature contains no systematic investigation about the effect of Si addition on the IGC.

We have studied that the effect of Si addition on the IGC of brazed Al–Mn–Cu alloy was investigated by electrochemical analysis and microstructure analysis9). In this study, detailed metallographic observations were particularly conducted to clarify the IGC mechanism of the alloy. Finally, we proposed the improvement of the IGC resistance of these alloys by Si addition.

2. Experimental Procedure

In this study, A3003 and Al-1.4 mass%Mn-0.3 mass%Fe-0.7 mass%Cu alloys with the Si content ranging from 0.1 to 1.5 mass% were used as specimens. The content of Mn and Cu in these alloys are greater than those of A3003. The samples were prepared by a conventional process which involves DC casting, homogenization, hot and cold rolling, and annealing. The final thickness was 0.5 mm and the temper type was O. The chemical composition of the alloys are listed in Table 1. The samples were heated to 600℃, maintained for 3 min, and then cooled under a high-purity N2 gas atmosphere to simulate a brazing treatment. Water-quenched samples and samples cooled at 50℃/min and 20℃/min were prepared to examine the effect of the post-brazing cooling rate on IGC susceptibility.

Table 1 Chemical compositions of the specimens (mass%).
Specimen Mn Cu Si Fe Al
A3003 0.91 0.15 0.10 0.28 bal.
0.1% Si 1.35 0.70 0.10 0.28 bal.
0.25% Si 1.34 0.70 0.26 0.28 bal.
0.5% Si 1.38 0.67 0.52 0.28 bal.
0.75% Si 1.34 0.67 0.77 0.29 bal.
1.0% Si 1.36 0.67 1.05 0.28 bal.
1.5% Si 1.34 0.71 1.58 0.29 bal.

IGC susceptibility was evaluated using anodic dissolution tests10) in aerated solution of 300 ppm Cl + 100 ppm SO42− at 25℃; the tests were conducted for 5 h at a current density of 1.0 mA/cm2.The samples were held at the potential which was nobler than the pitting potentials of both the grain boundary and grain interior. Then, all the components dissolve under same potential depending on the applied current. If each component has different pitting potential, the dissolution of less noble components only occurs practically. Therefore, IGC susceptibility can be evaluated by observation of dissolution morphology.

IGC occurs more easily in such anodic dissolution test than in an actual environment. In this study, it was defined that IGC susceptibility was high if the grain boundaries preferentially dissolved during the test. In contrast, the IGC susceptibility was considered to be low when the grain boundaries and the grain interiors dissolved concurrently. In addition, when the dissolution of grain boundaries was not observed, IGC susceptibility was judged as “none.”

The pitting potential was evaluated by anodic polarization measurement. A saturated calomel electrode was used as the reference electrode, and 2.67 mass% AlCl3 solution deaerated by bubbling with high-purity N2 was used at 40℃ as the electrolyte. Also, platinum electrode was used as a counter electrode. A sweep rate was 0.5 mV/s. Each specimen was masked except for the measurement area (10 × 10 mm2). The pretreatment consisted of etching in 5% NaOH solution at 50℃ for 30 s, desmutting in 30% HNO3 solution at room temperature for 60 s, and washing with tap water and deionized water. After the pretreatment, the anodic polarization measurement was performed. The measurement was completed when the current density increased to 50 mA/cm2. Additionally, a second measurement was performed after the specimen was etched for 5 s, desmutted for 30 s, and washed with tap water and deionized water. We speculated that the potential of the noble components could be measured in the second measurement because most of the less noble components on the measurement plane dissolved during the first measurement as a consequence of applying a high current. The pitting potential obtained from the first cycle was defined as that of the grain boundaries, whereas the pitting potential obtained from the second measurement was defined as that of the grain interiors.

The precipitation state at grain boundaries and in the grain interiors was observed by transmission electron microscopy (TEM, JEOL; JEM2010F) operated at an accelerating voltage of 200 kV, and by field-emission scanning electron microscopy (FE-SEM, Carl Zeiss; NVision). In addition, the formation state of the solute-depleted zone around the grain boundaries was investigated by aberration-correction scanning transmission electron microscopy (STEM, FEI; Titan G2 ChemiSTEM; operated at 200 kV) in conjunction with high-sensitivity energy-dispersive X-ray spectrometry (EDS). To detect the elemental distribution in the local regions around the grain boundaries, a 0.1-nm probe was used in EDS.

3. Results and Discussion

3.1 Anodic dissolution test results

Figure 1 shows the cross-sectional microstructures after the anodic dissolution tests. The water-quenched samples exhibited no IGC susceptibility. However, in the case of the samples cooled at 50℃/min or 20℃/min, IGC susceptibility increased as a whole. Additionally, more severe IGC occurred in 20℃/min cooled samples. This result clearly indicates that IGC occurred as a consequence of precipitation during cooling.

Fig. 1

Cross-sectional microstructures after the anodic dissolution tests.

To clarify the relation between IGC susceptibility and the content of Mn, Cu and Si, the IGC susceptibility of the samples cooled at 20℃/min or 50℃/min was ranked on a scale of 1 to 5. Rank 1 means that the grain interiors were dissolved without dissolution of grain boundaries. On the other hand, in the case of that only grain boundaries were dissolved, we judged as Rank 5. Furthermore, we defined as Rank 3 when the grain boundaries and the grain interiors were equally dissolved. The results are shown in Fig. 2. The 0.1% Si sample, which has high content of Mn and Cu, showed higher IGC susceptibility compared with A3003. Among the samples cooled at 20℃/min, IGC susceptibility was higher in those with low Si content, whereas IGC susceptibility was lower in those with Si content from 0.5 to 0.75 mass%. Furthermore, IGC susceptibility increased again when the Si content was higher than 1.0 mass%. These results reveal that adjusting the Mn/Si ratio to approximately 1.5 to 3.0 via Si addition can inhibit the IGC of the high-Mn, high-Cu alloy brazed and slowly cooled.

Fig. 2

Relation between IGC rank and the content of Si. IGC rank was judged from dissolution morphology. Rank 1: no dissolution of the grain boundaries. Rank 3 : equal level of dissolution between the grain boundaries and the grain interior. Rank 5: only dissolution of the grain boundaries.

3.2 Results of potential measurements on grain boundaries and grain interiors

Pitting potentials of the grain boundaries and the grain interiors are shown in Fig. 3. In the water-quenched samples with no IGC susceptibility, the potential of the grain boundaries equals to that of the grain interiors. In contrast, the samples cooled at 20℃/min, which are susceptible to IGC, exhibited a larger potential difference between the grain boundaries and the grain interiors compared to the water-quenched samples as a whole. These results suggest that the difference of the pitting potential between the grain boundaries and the grain interiors is suitable for evaluating the IGC. In the samples cooled at 20℃/min, A3003 has smaller potential difference than that of 0.1%Si sample. The potential difference decreased with increasing amount of Si added as a result of lowering the potential in the grain interiors; the potential difference increased again when the Si content exceeded 1.0 mass% because the potential in the grain interiors increased. This tendency is consistent with the effect of Si addition on IGC susceptibility (Fig. 1 and Fig. 2).

Fig. 3

Potential measurement results for the water-quenched samples and the samples cooled at 20℃/min. (b) and (c) indicate the anodic polarization curves of 0.1% Si sample cooled water-quenched and at 20℃/min. The pitting potential (Epit1) obtained from the first cycle was defined as that of the grain boundaries, whereas the pitting potential (Epit2) obtained from the second cycle was defined as that of the grain interiors.

The results show that the variation of IGC susceptibility corresponds with the change in the potential difference between the grain boundaries and the grain interiors with increasing amount of Si added. Furthermore, the potential difference appears to change in response to the variation of the potential of the grain interiors but not that of the potential of the grain boundaries. However, this interpretation based on Fig. 3 is incorrect because the potential of the water-quenched samples also changes with the amount of Si added. Also, the potential change of water-quenched samples may be depend on the amount of constituent particles formed during casting process, which are sample preparation process before brazing. Si addition gave that the potential became less noble due to decrease of solute Mn content because Al-(Mn, Fe)-Si constituent particles increased. In addition, the potential became noble because solute Si content increased due to more Si added.

As evident in Fig. 1, the lack of IGC in water-quenched samples clearly shows that the IGC occurs because of the precipitation during cooling. Therefore, we must consider the potential variation during cooling.

The pitting potential in the grain boundary and the grain interior were normalized by deducting the potentials of the water-quenched samples from those of the samples cooled at 20℃/min. Figure 4 shows the normalized pitting potentials as a function of the Si content of the sample. The normalized potential of the grain boundaries became noble with increasing Si content until the Si content reached 0.5 mass%. When the Si content exceeded 0.5 mass%, the normalized potential became less noble. Conversely, the normalized potential of the grain interiors became less noble with increasing Si content until the Si content reached 0.75 mass%. When the Si content was increased further, the normalized potential of the grain interiors became nobler. These variations in normalized potential result in a complicated change in the potential difference. Therefore, clarifying the reason for the change in IGC susceptibility with Si addition requires further investigation on the reason for the potential difference between grain boundary and grain interior.

Fig. 4

Results of potential measurements. The vertical left axis shows the normalized Epit and the right axis represents the potential difference between grain boundary and grain interior. Normalized Epit means the Epit of 20℃/min cooled sample deducted by the Epit of the water-quenched sample. ⊿E means the potential difference between grain boundary and grain interior.

3.3 Precipitation state of grain boundaries and grain interiors

Figure 5 shows the precipitation state on the grain boundaries and inside the grains of 0.1% Si, 0.75% Si, and 1.5% Si samples cooled at 20℃/min, as observed by TEM. Additionally, EDS and diffraction pattern analyses were carried out to identify the precipitates. In the water-quenched samples, few precipitates were observed on the grain boundaries, although a few Al–Mn system precipitates, which did not dissolve during the brazing treatment, were observed in the grains.

Fig. 5

TEM images (a) and SAED patterns (b) of the precipitation state of grain boundaries and grain interiors for the samples cooled at 20℃/min.

In the 0.1% Si sample, single-phase precipitation of Al6(Mn,Fe) and multi-phase precipitation of Al6(Mn,Fe) and CuAl2 were mainly observed. CuAl2 was precipitated along a grain boundary from the point where the Al6(Mn,Fe) intersected the grain boundary. To determine the formation mechanism of such unique multi-phase precipitates, we prepared a sample that was cooled from 600℃ to 400℃ at 20℃/min and was subsequently water-quenched. Observations for this sample revealed no multi-phase precipitation and only single-phase precipitation of Al6(Mn,Fe) on the grain boundaries. This result suggests that Al6(Mn,Fe) precipitated in the temperature range from 600℃ to 400℃. Subsequently, CuAl2 precipitated in the vicinity of Al6(Mn,Fe) precipitates. These precipitates (Al6(Mn,Fe)) likely functioned as nucleation sites at temperature below 400℃. In contrast, Al6(Mn,Fe) and Al15(Mn,Fe)3Si2 rarely precipitated in the grain interiors. The 0.75% Si sample contained more precipitates at grain boundaries compared to the 0.1% Si sample, and the precipitation morphology was also different. Single-phase precipitation of Al15(Mn,Fe)3Si2 was dominant, and no multi-phase precipitation was observed. In addition, numerous Al15(Mn,Fe)3Si2 precipitates were observed in the grain interiors because Si promotes the precipitation of Al–Mn system compounds11). Also, to investigate the change in the amount of Al–Mn system compounds precipitated during cooling, we plotted the difference in electrical conductivity between the sample water-quenched from 600℃ and the one cooled from 600℃ to 400℃ at 20℃/min and subsequently water-quenched as shown in Fig. 6. The difference in electrical conductivity was increased in the case of the 0.75% Si sample because the amount of precipitates increased. Also, the size of the Al–Mn system precipitates in both grain boundaries and grain interiors was finer than that of the Al–Mn system precipitates in the 0.1% Si sample. In the 1.5% Si sample, single-phase precipitates of Al15(Mn,Fe)3Si2 and coarse Si particles were observed on the grain boundaries. The precipitation state of the grain interiors in the 1.5%Si sample was approximately the same as that of the grain interiors in the 0.75% Si sample.

Fig. 6

Difference in electrical conductivity between the sample cooled at 20℃/min from 600 to 400℃ and the water-quenched sample.

To clarify the precipitation state on grain boundaries for samples cooled at 20℃/min with various Si content, we observed the precipitation state on a grain boundary with a total length of approximately 150 μm in each sample using FE-SEM and quantified the number of type of precipitates. The results are shown in Fig. 7. The 0.1% Si sample contained substantially more multi-phase precipitates compared to the 0.75% Si and the 1.5% Si samples. The total number of precipitates on the observed grain boundaries increased for the 0.75% Si and the 1.5% Si samples. No multi-phase precipitates were observed, although coarse Si particles were present in the 1.5% Si sample.

Fig. 7

Quantitative evaluation results for the samples cooled at 20℃/min related to the precipitation state on grain boundaries. The type of precipitate was classified as below. (a) Single-phase precipitation of Mn bearing precipitate, (b) Multi-phase precipitation of Mn bearing precipitate and CuAl2, (c) Single-phase of CuAl2, (d) Single phase of Si particle.

As previously mentioned, the precipitation state on the grain boundaries and in the grain interiors changed with the Si content. IGC susceptibility may varied with the Si content because the change in the potential difference between the grain boundaries and the grain interiors is resulted from the changes in element distribution. Consequently, we presumed that the sample with high IGC susceptibility had a solute-depleted zone around the grain boundaries. Therefore, EDS mapping analysis near a grain boundary was performed.

3.4 Solute-depleted zone near grain boundaries

The solute distribution near the grain boundaries was investigated by STEM-EDS and the line analysis results are shown in Fig. 8 and Fig. 9, respectively. The line profiles in Fig. 9 were extracted from the mapping data, and the vertical axis values were indicated as arbitrary units. In the 0.1% Si sample, Mn/Cu depleted zones with a width of 200 nm were formed continuously in the vicinity of grain boundaries because of preferential precipitation of Al6(Mn,Fe) and CuAl2 on the grain boundaries. In contrast, in the 0.75% Si sample, no Mn-depleted zone was formed because the Mn content of the grain interiors became the same as that of the grain boundaries as a result of a decrease of the Mn content in the grain interiors. This depletion of Mn in the grain interiors was a consequence of enhanced precipitation of Al15(Mn,Fe)3Si2 in the grain interiors despite the decrease in Mn content caused by increasing precipitation of Al15(Mn,Fe)3Si2 on the grain boundaries. Furthermore, Cu-depleted zone was not formed because CuAl2 did not preferentially precipitate on the grain boundaries. In the 1.5% Si sample, Mn/Cu-depleted zone was not formed as with the 0.75% Si sample. However, a 2000-nm-wide Si-depleted zone formed as a result of the precipitation of coarse Si particles on the grain boundaries. Mn, Cu, and Si raise the potential in Al when these elements are dissolved in solid solution12). If the Mn/Cu/Si-depleted zone forms in the vicinity of grain boundaries, the potential of the grain boundaries becomes less noble than that of the grain interiors.

Fig. 8

STEM-EDS elemental mappings at the vicinity of a grain boundary in the Si bearing sample cooled at 20℃/min, as obtained by STEM-EDS analysis.

Fig. 9

Line-profile-extracted elemental maps near grain boundaries of the Si bearing samples cooled at 20℃/min, as investigated using STEM: (a) 0.1% Si, (b) 0.75% Si, (c) 1.5% Si, and (d) 1.5% Si. (a), (b), and (d) represent the results at the vicinity of a grain boundary, whereas (c) represents the result over a wide area near the grain boundary.

Also, in all of the samples, Cu segregation in the immediate vicinity (within several tens of nanometers) of grain boundaries was observed. This segregation may result in the formation of the Cu-depleted zone. However, the formation of the Cu depleted zone was not caused by Cu segregation on grain boundaries as already showed that the 0.75% Si and 1.5%Si samples, which have no Cu-depleted zone, also exhibit Cu segregation on grain boundaries. Additionally, Cu segregation on grain boundaries was observed in the 0.1% Si sample that was water-quenched from 600℃ (Fig. 10). Cu, which has large misfit to Al13), may segregate on grain boundaries because such segregation lowers the system-wide energy, like solute segregation on a dislocation core. Similar grain-boundary segregation at the early stage of aging was reported for Al–Cu alloys14). Also, Cu segregation may promote the formation of a Cu-depleted zone due to accelerating grain boundary precipitation of CuAl2. The acceleration is caused by the increase of the driving force for grain boundary precipitation. This effect will be the subject of a future study.

Fig. 10

Line profiles of Cu, Mn and Si near the grain boundary of the water-quenched 0.1% Si sample.

4. IGC Mechanism of Al–Mn–Cu–Si System Alloy

From the aforementioned results, we estimated the IGC mechanism of the brazed Al–Mn–Cu–Si alloy. The mechanism is shown schematically in Fig. 11.

Fig. 11

The estimated IGC mechanism of brazed Al–Mn–Cu–Si alloy. “SDZ” indicates solute-depleted zone.

In the low-Si-containing alloy (0.1% Si sample), Al6(Mn,Fe) precipitated at the temperature range from 600℃ to 400℃ during cooling after brazing. These precipitates preferentially formed on the grain boundaries. Afterwards, the Mn-depleted zone was formed in the vicinity of the grain boundaries. Furthermore, CuAl2 precipitated on the grain boundaries at temperature below 400℃. We propose that the Al6(Mn,Fe) precipitates on the grain boundaries functioned as nucleation sites for the precipitation of CuAl2. Consequently, the Cu-depleted zone was also formed near the grain boundaries. Hence, potential of grain boundary became much less noble than that of grain interior and IGC easily occurred.

In the optimum-Si-containing alloy (0.75% Si sample), Al15(Mn,Fe)3Si2 precipitated at temperature above 400℃. The amount of the precipitates was larger than that in the low-Si-containing alloy. The precipitation occurred at both the grain boundaries and the grain interiors because added Si promoted precipitation, and Mn content in grain interior decreased due to enhanced precipitation of Al15(Mn,Fe)3Si2 in the grain. Therefore, the formation of a Mn-depleted zone was inhibited. The Cu-depleted zone was not formed given the absence of CuAl2 precipitates on the grain boundaries at temperatures below 400℃ in this alloy. The potential of grain boundary was almost same as that of grain interior and IGC hardly occurred.

In excess-Si-containing alloy (the 1.5% Si sample), the formation of the Mn-depleted zone was inhibited because of the interception of the preferential precipitation of Al15(Mn,Fe)3Si2 on the grain boundaries, as in the case of the alloy containing moderate Si. Because of the absence of CuAl2 precipitates on the grain boundaries, a Cu-depleted zone was not formed. However, coarse Si particles precipitated on the grain boundaries because the 1.5% Si sample contained supersaturated Si. Hence, Si-depleted zone was formed nearby grain boundaries. As a result of this Si-depleted zone, IGC susceptibility was high due to much less noble potential of grain boundary.

5. Conclusions

In this paper, the effect of Si addition on IGC susceptibility of brazed Al–Mn–Cu alloy was studied, and electrochemical analysis and metallographic observations were carried out. On the basis of the obtained results, we proposed a mechanism of IGC. The following conclusions were drawn:

(1) Al-1.4 mass%Mn-0.7 mass%Cu-0.1 mass%Si alloy exhibits higher IGC susceptibility than A3003. Actually, IGC occurred for the former with 50℃/min cooling rate although no IGC was observed for A3003 cooled at a rate slower than 20℃/min.

(2) In the case of Al-1.4 mass%Mn-0.7 mass%Cu alloys, IGC susceptibility drastically depends on the Si content. The alloys with low or high Si content exhibited high IGC susceptibility. In contrast, the alloy with an optimum Si content inhibited IGC.

(3) Adjusting the Mn/Si ratio to approximately 1.5 to 3.0 via Si addition can inhibit IGC of the high-Mn, high-Cu alloy brazed and slowly cooled.

(4) IGC susceptibility depends on the difference of pitting potentials between the grain boundaries and the grain interiors. The behavior is resulted from precipitation state and the formation of solute-depleted zone near the grain boundaries. A large potential difference results in an increase of IGC susceptibility.

(5) In the low-Si-containing alloy, the Mn/Cu-depleted zone is formed by the preferential precipitation of Al6(Mn,Fe) and CuAl2 on the grain boundaries. Al6(Mn,Fe) likely works as a nucleation site for the precipitation of CuAl2.

(6) In the optimum-Si-containing alloy, no Mn-depleted zone is formed because the decrease of the Mn content in the grain interiors due to enhanced precipitation of Al15(Mn,Fe)3Si2 in the grain. Additionally, Cu-depleted zone is also not formed because preferential precipitation of CuAl2 on the grain boundaries is prevented.

(7) In the excess-Si-containing alloy, Mn/Cu-depleted zones are not formed as with the optimum-Si-containing alloy. However, Si-depleted zone is formed, as indicated by the preferential precipitation of coarse Si particles on the grain boundaries.

REFERENCES
 
© 2017 The Japan Institute of Light Metals
feedback
Top