4.1 Why did not type IV creep cracking take place?
Type IV creep cracking is a kind of common premature failure in MHRS welded joint during service, and it is well established that the type IV creep cracking happens under the interaction between high temperature and low stress and is located in FGHAZ or ICHAZ. Up to now, much in-depth research has been conducted on its mechanism. Wang et al.[14, 15] indicated that type IV creep cracking nucleated in ICHAZ of P91 welded joint, and was the result of Cr segregation in grain-scale. Pandey C et al.[16] investigated the effect of PWHT on type IV creep cracking of P91 welded joint. They pointed out that although PWHT (760 ℃ + 2 h) could mitigate hardness unevenness throughout the welded joint, high temperature and pressure would bring it out again during service. They also unraveled that Post Welding Normalizing Treatment (PWNT) could basically eradicate the tendency of type IV creep cracking in P91 welded joint. Wang et al.[17] reported that the type IV creep cracking in P92 welded joint nucleated at the region with relatively high temperature (a little higher than Ac3, namely FGHAZ) and under low stress (lower than 120 MPa), and coarse Laves phase at grain boundaries, loss of lath morphology and low hardness were all stimuli to type IV creep cracking. All the research on type IV creep cracking above is focused on the 2nd and 3rd generation of 9Cr MHRS. As the representative figure of the 4th generation of 9Cr MHRS, MarBN steel is quite different from P91 and P92 in chemical composition, so its microstructure evolution during the welding process would be also quite different.
Recently, Japanese researchers[18] found out that when B content reached 100 ppm in P93 steel, diffusive phase transformation in HAZ could be suppressed, so M → A transformation in the heating process would take place in the displacive form which could impede the formation of FGHAZ and ICHAZ. On the other hand, traditional FGHAZ or ICHAZ was the prerequisite for type IV creep cracking, so MHRS with high B content exhibited a low type IV creep cracking tendency. Based on the findings above, researchers[19, 20] further pointed out that grain refinement in CGHAZ might result from the formation of so called “lath austenite[21]” in the heating process. Not only would this “lath austenite” maintain certain orientation relationship with the original martensite, but it also retained the sub-structure of the martensite (lath and dislocation density) which could increase the stored energy. During subsequent heating process, because of the stored energy it inherited, the “lath martensite” would recrystallize, which could eventually lead to grain refinement.
According to the results of this research, another mechanism of why type IV creep cracking did not take place was put forward. As could be seen in Fig. 5d, although FGHAZ had inherited the coarse martensitic morphology from BM, some areas (the dark “fine grain structure”) still showed the sigh of diffusive M→A transformation. Therefore, high B content (0.012% in wt. in this research) could not suppress diffusive M→A transformation completely, which meant the diffusive M→A transformation might still happen in MarBN if higher driving force (higher temperature) for phase transformation could be reached. On the other hand, obvious grain refinement was observed in CGHAZ (Fig. 5b). As mentioned above, some researchers ascribed the grain refinement in CGHAZ to recrystalization, but there was another possibility: With temperature increasing, the driving force would be enough to make diffusive M → A transformation take place completely in CGHAZ, thereby refining the grains. Firstly, the as-delivered state of MarBN was normalizing + tempering. It was well known that tempering could basically eliminate the residual stress caused by normalizing and significantly reduce the dislocation density in martensitic laths simultaneously, so stored energy stemming from the dislocation retained in “lath austenite” might not be adequate to make recrystalization possible. Secondly, B content in this experimental MarBN could not suppress diffusive M→A transformation completely. As for the reason why MarBN welded joint fractured at WM rather than FGHAZ or ICHAZ (type IV creep cracking) below 220 MPa, it might be caused by the strength reduction in WM (compare Fig. 6 and Fig. 10). In other words, if WM could maintain high strength during creep, type IV creep cracking was still likely to take place in MarBN steel welded joint.
4.2 Why did fracture happen at WM below 220 MPa?
For the as-delivered state of MarBN steel welded joint, the hardness of WM was higher than that of BM (Fig. 6). In the process of high stress creep (220MPa), there was no obvious degradation taking place in WM because of short rupture time. Thus the strength of WM was still higher than that of BM, thereby making the fracture happen in BM. But in the condition of low stress creep (below 220MPa), the rupture time was long and therefore both WM and BM would experience evident degradation such as M23C6 coarsening, dislocation recovery and sub-grain growth. Nevertheless, the degradation in WM was more serious than that in BM (Fig. 9a and 9b), thereby leading to a faster hardness reduction in WM, so the fracture would happen in WM under this circumstance. Abe[22] pointed out that when B content was more than 100 ppm in 9Cr-3W-3Co, the coarsening of M23C6 could be suppressed. Actually, in most cases, B content in WM was much lower than that in BM because of burning loss during welding process, and it was just only a half of B content in BM in this research (60 ppm in WM versus 120 ppm in BM according to Table 1). This would lead to a faster M23C6 coarsening rate in WM than that in BM, which meant more serious degradation in WM. This could also be proved by a more significant hardness reduction in WM compared with BM during creep (see Fig. 6 and Fig. 10 for comparison).