*3.3. Microstructure of Corrosion Attacked Surface*

Figures 6 and 7 show the metallographic cross sections of the CR20 and CR50 alloys after the IGC test. To explore the IGC behavior, the specimens were etched using Keller's reagent for 10 s. Figures 6a and 7a indicate the estimated corrosion depths after 0 h of sensitization. The depths are 16.75 μm and 17.71 μm for the CR20 and CR50 alloys, respectively. The penetration depth increases with increasing annealing time. The precipitation of β at grain boundaries increases with increasing sensitization time. It is assumed that the anodic precipitation of the CR50 alloy is deeper than in the CR20 alloy, which results in the IGC rate being higher in this alloy. As shown in Figures 6c and 7c, the grains start to be detached from the specimen after 48 h at 100 ◦C.

**Figure 6.** The surface of the CR20 alloy after NAMLT at different sensitization times: (**a**) 0 h, (**b**) 7 h, (**c**) 48 h, (**d**) 144 h, and (**e**) 207 h.

Figure 8 shows the maximum corrosion depth of both CR20 and CR50 alloys. The values are similar at the maximum sensitization time. On the other hand, the CR20 alloy shows a higher mass-loss rate compared to the CR50 alloy (Figure 5).

The continuity of β precipitation at grain boundary is critical for the IGC depth. The grain size, on the other hand, is a more important factor in the mass-loss rate at the early periods of sensitization. While there is not a significant difference between the alloys in maximum IGC depth, the NAMLT results show a higher mass-loss rate in the CR20 alloy after the long heat treatment time. This means that the CR20 alloy is more susceptible to IGC.

**Figure 7.** The surface of the CR50 alloy after NAMLT at different sensitization times: (**a**) 0 h, (**b**) 7 h, (**c**) 48 h, (**d**) 144 h, and (**e**) 207 h.

**Figure 8.** Maximum IGC depths of CR20 and CR50 alloys.

#### *3.4. Schematic of β Phase Distribution in Grain Boundary*

Figure 9 shows the SEM images of the CR20 and CR50 alloy surface after sensitization. The grain boundary is covered with β precipitation. The specimens were etched using H3PO4 etchant to selectively reveal the grain boundary covered by the anodic β precipitation. The continuity of β precipitation can be indirectly observed by the microstructure of the etched surface. At sensitization time of 7 h, the grain boundaries were more discernible in the CR50 alloy. This indicates that the CR50 alloy is more susceptible to IGC at 7 h sensitization. Figures 10 and 11 show a close look-up at the microstructure of the etched

surface. The grain boundary is covered by β precipitation. The β precipitation is more discontinuous in the CR20 alloy.

**Figure 9.** Microstructure of sensitized CR20 and CR50 alloys.

**Figure 10.** Close inspection of β precipitation covered grain boundary: (**a**) CR20 alloy, Sensi. 48 h, (**b**) CR50 alloy, Sensi. 48 h.

**Figure 11.** Close inspection of β precipitation covered grain boundary: (**a**) CR20 alloy, Sensi. 207 h, (**b**) CR50 alloy, Sensi. 207 h.

Previous studies found that the precipitate growth rates increased with rolling reduction [18,19]. A high density of dislocations can lower the activation energy, which most likely initiates the precipitation in the rolled specimen [18]. Increasing the dislocation

density results in enhancing the diffusivity of Mg atoms at sensitization treatment due to pipeline diffusion [19]. On the other hand, there is no difference in the continuity of the precipitation at grain boundary at long sensitization times (Figure 11) by precipitation [18]. Considering both mass-loss results and maximum IGC depths (Figures 5 and 9), the grain size thus becomes a crucial factor in IGC at longer sensitization times.

#### *3.5. Direct Observation of β Precipitation Distribution*

Figure 12 shows the thickness of the precipitates for the CR20 and CR50 alloys. The size of the precipitates was calculated from the crossline thickness of the precipitates by TEM image (Figure 13). Figure 13e represents the TEM image of the CR20 alloy with EDS mapping of magnesium at 48 h of sensitization. The thickness was found to be 6.1 ± 1.7 nm for the CR20 alloy and 7.5 ± 3.0 nm for the CR50 alloy at 48 h of sensitization time. The precipitate thickness at 207 h is higher. The thickness of β-precipitates is crucial for the IGC rate as it affects the continuity of the precipitation at the grain boundary. Figure 13a reveals that the β precipitation of the CR20 alloy in the early period of sensitization is discontinuous. This results in a superior corrosion resistance compared to the CR50 alloy at the early sensitization time. The size of β precipitates at longer sensitization times becomes larger for the CR20 alloy. Both Figure 13c,d show that β precipitation was almost continuously distributed at the grain boundary with almost the same thickness as shown in Figure 12. Zhang et al. also agree that the kinetics of precipitation growth is reduced with sensitization time [28]. The results of this study show that the IGC significantly depended on the grain size for long-term sensitization, as compared to the size of precipitates.

**Figure 12.** Thickness of β-phase precipitates in the CR20, and CR50 alloys sensitized for 48 and 207 h, respectively.

**Figure 13.** TEM of β precipitates in: (**a**) CR20 alloy, 48 h; (**b**) CR50 alloy, 48 h; (**c**) CR20 alloy, 207 h; (**d**) CR50 alloy, 207 h; and (**e**) TEM-EDS (Mg element) CR20 alloy, 48 h.

#### **4. Discussion**

In this research, we found that both grain size and continuity of β precipitation at grain boundaries are important factors affecting the Al-Mg IGC susceptibility.

The TEM image in Figure 13 shows that the β precipitates are much thicker in the CR50 alloy at an early period of sensitization. The β precipitates thickness, however, is almost the same at long-term sensitization. Some researchers suggested that grain boundary misorientation is a crucial factor for the growth rate and the final size of β precipitation. These factors affect the continuity of β precipitation at grain boundaries [10,12,26,29–34]. Wang et al. concluded that some grain boundaries, e.g., low-angle grain boundaries generated by plastic deformation, are not susceptible to IGC [28]. On the other hand, D'Antuono reported that although the β precipitation was preferentially formed at low-angle grain boundary, the final size of precipitation was larger at high-angle grain boundary [18]. The influence of grain boundary plane orientation was reported to affect the continuity of precipitation. It was found that grain boundary (GB) planes close to {110} direction facilitate the β precipitation while the GB plane near {100} direction may be resistant to β precipitation [31,32].

Previous studies showed that the rolled specimen had a high resistance to IGC coming from the confluence of refined grain size and the fraction of low-angle grain boundaries [17,25,26]. In this study, it was revealed that the effect of grain size on IGC needs to be considered depending on the sensitization heat treatment which affects the formation of anodic β-Mg2Al3 precipitation at the grain boundary. High dislocation density induced by cold rolling facilitates the precipitate growth rates. The formation of anodic β-Mg2Al3 is affected by temperature and the presence of prior strain [35,36]. The increased dislocation density tends to lower the nucleation temperature and reduce Mg diffusion at a lower temperature [18,19]. These factors are reflected in increasing the susceptibility of the CR50 alloy in the early period of sensitization. In this situation, the dislocation density and grain boundary type affect the IGC susceptibility more significantly compared to the grain size. On the other hand, the grain size affects the IGC susceptibility of cold-rolled Al-6Mg alloy more dramatically than the grain boundary type. It was found that the large-grained material tends to be more susceptible to IGC when the precipitation is continuously formed at the grain boundary due to sufficient sensitization time.
