Constant‐load creep‐rupture tests were performed on single Si‐C‐O fibers (Nicalon). Test environments included pure carbon monoxide (CO), pure argon gas, and a mixture of CO and argon gas with a CO partial pressure of 40 kPa. Fibers were tested at temperatures of 1200°‐1400°C and at nominal applied stresses of 0.15–0.7 GPa. The as‐received and crept specimens were characterized by means of scanning electron microscopy, transmission electron microscopy, X‐ray photoemission spectroscopy, electron‐probe microanalysis, Auger electron spectroscopy, and thermo‐gravimetric analysis. In pure argon, the microstructure of the Nicalon fiber was unstable, which was attributed to the decomposition of the silicon oxycarbide phase, which resulted in CO and silicon monoxide gas evolution and silicon carbide grain growth. Fiber shrinkage was observed at temperatures <1300°C at low applied stresses. At high stresses, fibers exhibited only primary creep. In the CO/ argon‐gas environment, very limited grain growth and a smooth carbon coating were observed at the fiber surface at temperatures <1350°C. At all applied stresses, fibers exhibited steady‐state creep whose rates, strains, and times to failure were higher than those observed in argon. The apparent activation energy for creep of Nicalon fibers in the CO/argon‐gas environment was 435 kj/mol. At temperatures >1350°C in the CO/argon‐gas environment, however, the fiber behaved as in pure argon. Tests in pure CO only resulted in lower strains to failure and thicker carbon layers on the fiber surface. A rheological model based on the viscous flow of a concentrated suspension was proposed to describe the fiber deformation. The continuously decreasing creep rate in argon was suggested to be related to the continuous increase of the total solid volume fraction, which affects the fiber viscosity. On the other hand, the steady‐state creep of Nicalon with a stable microstructure in the CO/ argon‐gas environment was characterized by a Newtonian‐type viscous flow, which supports the predictions of the model.
Read full abstract