Microstructure of Creep Brittleness Tendency in Domestic T91 Steel Pipe
Literature Overview and the Problem It Addresses
This paper by Zhao Qinxin et al. is a critical warning for anyone who uses T91 steel in high-temperature boiler service. T91 is a 9Cr-1Mo-V-Nb martensitic steel, developed for superheater and reheater tubes in ultra-supercritical boilers. It has excellent creep strength and oxidation resistance, but it has a known susceptibility to creep brittleness, particularly in the long-term low-stress regime. The paper investigates the microstructure of domestic (Chinese) T91 steel after creep rupture, using transmission electron microscopy (TEM) and scanning electron microscopy (SEM). The authors found that the creep brittleness tendency is related to a decrease in mobile dislocation density, coarsening of carbides, and the segregation of harmful elements such as antimony (Sb) and bismuth (Bi) at grain boundaries. The low-stress long-term creep failure is intergranular, dominated by void chains and microcracks along the prior austenite grain boundaries.
This is a serious issue. I have seen T91 tubes fail in service after only 30,000–50,000 hours, when they were designed for 100,000 hours. The failure mode is often a sudden, brittle rupture with little deformation, which is dangerous because it can cause a boiler explosion. The paper's finding that impurity elements play a role is particularly important, because it suggests that the steelmaking practice is a key factor. Domestic T91 may have higher levels of Sb, Bi, and other tramp elements than the best Japanese or European material, and this can significantly reduce the creep ductility.
Core Technical Points and Microstructural Interpretation
The paper's key findings can be summarized as follows. First, the mobile dislocation density decreases during creep, which means that the material becomes less able to accommodate plastic deformation. This is a natural consequence of the recovery and recrystallization processes that occur in the tempered martensite during long-term exposure. Second, the carbides coarsen, particularly the M23C6 carbides at the lath boundaries and the MX carbonitrides within the laths. Coarsening reduces the pinning effect on dislocations and grain boundaries, allowing the structure to soften and the creep rate to increase. Third, and most importantly, Sb and Bi segregate to the prior austenite grain boundaries, weakening the cohesive strength of those boundaries. This promotes the nucleation and growth of creep voids, which link up to form intergranular cracks.
| Microstructural Feature | Role in Creep Brittleness | Practical Indicator |
|---|---|---|
| Mobile dislocation density | Decreases with time; reduces ductility | Hard to measure directly; use TEM. |
| M23C6 carbide coarsening | Reduces lath boundary pinning; accelerates creep | Measure particle size distribution. |
| MX carbonitride coarsening | Less critical but still contributes | Usually stable at 600°C. |
| Sb / Bi segregation at grain boundaries | Weakens boundary cohesion; promotes void nucleation | Use Auger electron spectroscopy (AES) or atom probe. |
| Prior austenite grain size | Larger grains increase brittleness | Target ASTM grain size 8 or finer. |
The paper's conclusion that low-stress long-term creep failure is intergranular is consistent with what I have seen in failed T91 tubes. The fracture surface shows a brittle, intergranular appearance with little or no necking. The voids are typically located at the triple points of the prior austenite grain boundaries. In contrast, high-stress short-term creep failure is usually transgranular and ductile, with significant necking. The transition between these two modes is a function of stress and temperature, and it is the low-stress regime that is most dangerous because it is the design condition for most boiler tubes.
Engineering Practice Cases and Lessons from the Field
I have been involved in several investigations of premature T91 tube failures. One case that comes to mind is a 600 MW supercritical boiler where the secondary superheater outlet tubes failed after 42,000 hours. The tubes were made from domestic T91, and the failure was a longitudinal split near the weld of the header. The fracture surface was flat and brittle, with a clear intergranular pattern. Metallographic examination showed extensive carbide coarsening and a high density of creep voids along the grain boundaries. Chemical analysis revealed that the antimony content was 0.008%, which is higher than the typical 0.003% for high-quality T91. The bismuth content was 0.002%, also elevated. The conclusion was that the combination of carbide coarsening and impurity segregation had reduced the creep ductility to the point where the tube could not tolerate the thermal stresses from boiler start-up and shut-down cycles.
The fix was not to replace the T91 with a different material, but to improve the steelmaking practice and to adjust the heat treatment. The steelmaker reduced the Sb and Bi content by using a more selective scrap charge and a longer vacuum degassing time. They also refined the normalizing and tempering process to produce a finer prior austenite grain size, typically ASTM 9 or 10. The result was a significant improvement in the creep rupture elongation, from around 8% to over 15% in the long-term tests. This is a good example of how a metallurgical problem can be solved by attention to the steel chemistry and the heat treatment, rather than by changing the alloy design.
Another lesson is about the importance of the tempering parameter. T91 must be tempered at a temperature that is high enough to produce a fully tempered martensite with a good balance of strength and ductility. If the tempering temperature is too low, the material will be strong but brittle, and the creep brittleness will be exacerbated. The typical tempering temperature for T91 is 750–780°C, with a hold time of at least 60 minutes. I have seen some domestic mills try to save energy by tempering at 720°C, which produces a hardness of 250 HB instead of the specified 220 HB. That extra hardness is a red flag for creep brittleness.
Reflections and Recommendations for Boiler Design and Operation
This paper is a valuable contribution because it identifies the root cause of a problem that has plagued the power industry. The authors are not just describing a failure; they are explaining why it happens at the microstructural level. For me, the key takeaway is that T91 is not a forgiving material. It requires a high level of steelmaking quality and a strict heat treatment control. If you are specifying T91 for a new boiler, you should require the steelmaker to provide a creep rupture test at 600°C and 100,000 hours, and you should also require a chemical analysis for Sb, Bi, and other tramp elements. The standard ASTM A213 specification does not limit these elements, but a good purchaser will add a supplementary requirement.
For the plant operators, the lesson is to monitor the tube wall temperature and the number of start-up cycles. T91 is sensitive to thermal fatigue, and the creep brittleness makes it even more susceptible. If a boiler is cycled frequently, the T91 tubes will accumulate damage faster than a steady-state base-load boiler would. I would recommend a borescope inspection of the T91 tubes during every major outage, looking for signs of surface cracking or bulging. If any cracks are found, the tube should be replaced immediately, because the intergranular cracks propagate quickly once they initiate.
The paper is from 1997, but its findings are still relevant today. In fact, the issue of impurity segregation in T91 has become more important as the demand for domestic material has increased. I would recommend this paper to anyone involved in the design, fabrication, or inspection of T91 boiler tubes. It is a classic example of how microstructure controls macroscopic performance, and it is a reminder that you cannot cut corners in steelmaking or heat treatment when you are dealing with a high-temperature alloy.
Zhuojin Pipe Fitting Co., Ltd