J M. Ehrman et al Composites Science and Technology 67(2007)1425-1438 of the Si species has occurred in tested specimens, and sug- fatigue lives for stresses <125 MPa in air (100 MPa in gest that such migration occurs at a faster rate during test- steam), and at least two orders of magnitude shorter than Ing in steam. fatigue lifetimes for the stresses >125 MPa in air The EDS revealed deposition of Si in the matrix of the (100 MPa in steam). In air, introducing a hold period of omposite specimens subjected to prior testing at 1200C up to 100-s duration into a fatigue cycle has no effect on in steam environment, suggesting that the loss of mullite cyclic life for the maximum stress levels <125 MPa. Run- from the fiber may be the mechanism behind the degraded out was achieved in all 125 MPa cyclic tests with hold time creep performance in steam. Alternatively, it is recognized The run-out specimens retained 100% of tensile strength that a stress-corrosion mechanism may be the cause of For the stress levels >154 MPa, superposition of a 10-9 poor creep resistance and fast fracture observed at elevated hold causes a significant reduction in fatigue life. Increas- temperature in steam. Earlier studies [39-44]suggested that ing the hold period to 100 s further reduces fatigue life by static fatigue of glass was a chemical process, where crack an order of magnitude In steam for the fatigue stress levels growth resulted from and was controlled by a stress- >100 MPa, a superposition of a hold time of at least 10-s enhanced chemical reaction between the glass and water duration onto a fatigue cycle dramatically degrades cyclic in the environment. Michalske and Freiman[45] proposed life, reducing it to the much shorter creep life at a given a qualitative chemical model to describe the interaction of applied stress. Increasing the hold period does not cause water with mechanically strained Si-o-Si bond bridging further degradation in cyclic life In air, lowest strains are the crack tip Michalske and Bunker [46, 47]examined the accumulated in fatigue and highest strains, in creep. Strains role of mechanical strain in accelerating chemical reactions accumulated in fatigue tests with hold fall in the intermedi- between the si-o bonds at the crack tip and environmental ate range. Strain accumulations increase with increasing molecules, While in the absence of strain the si-o bond hold time. In steam evolution of maximum strain in fati remained relatively inert to water and other chemicals gue with hold is akin to that in creep. For a given maxi- known to cause stress corrosion in silica glass, the highly mum stress, strains accumulated in cyclic tests with hold strained Si-o bonds reacted with water at least 8 orders are close to those accumulated in creep and significantly of magnitude faster than the unstrained bonds. Michalske higher than those accumulated in fatigue Steady-state rate and Bunker [46, 47] developed a quantitative chemical- of strain accumulation was reached in all tests. Presence of kinetics-based model to predict the rate of crack growth steam considerably accelerates strain accumulation under in silica glass in humid conditions as a function of the both cyclic and static loadings. In air, the rate of strain applied stress. This model describes a fracture rate law in accumulation increases with hold time at maximum stress which the crack growth rate increases exponentially with In steam, strain rates in cyclic tests with hold of any dura- the applied stress intensity tion are close to those obtained in creep and significantly Wannaparhun [37] and Campbell [38] reported strength higher than those observed in fatigue degradation after a no-load exposure of the composite material to water vapor at elevated temperatures. In con- 42. Cyclic-static and static-cyclic block loadings trast, the behavior observed in the current work is strongly influenced by the loading conditions. Note that strength Effects of prior fatigue on creep performance of the losses similar to those observed by Campbell et al. [38]after N720/A composite were studied in cyclic-static block load 1000 h of no-load exposure were seen after only 28 h(10. ing tests at 1200C for maximum stress levels of 125 and cycles at 1 Hz) of fatigue cycling in steam at 1200C. The 154 MPa in laboratory air, and for maximum stress levels mechanism behind the degraded performance in steam of 100 and 125 MPa in steam. The fatigue block contained observed in this study as well as in prior work [27] is 10 cycles In air, prior fatigue dramatically improves creep strongly dependent on the applied load, and may indeed resistance of the CMC, increasing creep life by one to two be a stress-corrosion mechanism involving a chemical orders of magnitude. Creep strain rates for the pre-fatigued tion of moisture with strained si-o bonds at the crack specimens are at east an ord rder of magnitude lower than those for the as-processed material. For the pre-fatigued 4. Concluding remarks composite the run-out stress was 125 MPa, while the as- processed composite did not achieve a run-out. The pre- 4.. Fatigue with hold times fatigued specimen that achieved a run-out retained over 100% of its tensile strength and N76% of its modulus. In The effect of hold times at maximum stress on fatigue steam, prior fatigue reduced creep resistance of the behavior of the N720/A ceramic composite was investi- composite gated at 1200C for maximum stress levels of 125 and Efects of prior creep on fatigue durability of the N720/ 154 MPa in laboratory air, and for maximum stress levels A composite were studied in static-cyclic block loading of 100 and 125 MPa in steam hold times were 10 and tests at 1200 oc for the maximum stresses of 125 and 100 s. In both air and steam environments, static loading 100 MPa in air and in steam, respectively. In air the creep is considerably more damaging than cyclic loading Creep period was 2 h and in steam, 0.75 h In air, prior creep did lifetimes are at least one order of magnitude shorter than not affect fatigue life up to 10 cycles, the fatigue run-outof the Si species has occurred in tested specimens, and sug￾gest that such migration occurs at a faster rate during test￾ing in steam. The EDS revealed deposition of Si in the matrix of the composite specimens subjected to prior testing at 1200 C in steam environment, suggesting that the loss of mullite from the fiber may be the mechanism behind the degraded creep performance in steam. Alternatively, it is recognized that a stress-corrosion mechanism may be the cause of poor creep resistance and fast fracture observed at elevated temperature in steam. Earlier studies [39–44] suggested that static fatigue of glass was a chemical process, where crack growth resulted from and was controlled by a stress￾enhanced chemical reaction between the glass and water in the environment. Michalske and Freiman [45] proposed a qualitative chemical model to describe the interaction of water with mechanically strained Si–O–Si bond bridging the crack tip. Michalske and Bunker [46,47] examined the role of mechanical strain in accelerating chemical reactions between the Si–O bonds at the crack tip and environmental molecules. While in the absence of strain the Si–O bond remained relatively inert to water and other chemicals known to cause stress corrosion in silica glass, the highly strained Si–O bonds reacted with water at least 8 orders of magnitude faster than the unstrained bonds. Michalske and Bunker [46,47] developed a quantitative chemical￾kinetics-based model to predict the rate of crack growth in silica glass in humid conditions as a function of the applied stress. This model describes a fracture rate law in which the crack growth rate increases exponentially with the applied stress intensity. Wannaparhun [37] and Campbell [38] reported strength degradation after a no-load exposure of the composite material to water vapor at elevated temperatures. In con￾trast, the behavior observed in the current work is strongly influenced by the loading conditions. Note that strength losses similar to those observed by Campbell et al. [38] after 1000 h of no-load exposure were seen after only 28 h (105 cycles at 1 Hz) of fatigue cycling in steam at 1200 C. The mechanism behind the degraded performance in steam observed in this study as well as in prior work [27] is strongly dependent on the applied load, and may indeed be a stress-corrosion mechanism involving a chemical reac￾tion of moisture with strained Si–O bonds at the crack tip. 4. Concluding remarks 4.1. Fatigue with hold times The effect of hold times at maximum stress on fatigue behavior of the N720/A ceramic composite was investi￾gated at 1200 C for maximum stress levels of 125 and 154 MPa in laboratory air, and for maximum stress levels of 100 and 125 MPa in steam. Hold times were 10 and 100 s. In both air and steam environments, static loading is considerably more damaging than cyclic loading. Creep lifetimes are at least one order of magnitude shorter than fatigue lives for stresses 6125 MPa in air (100 MPa in steam), and at least two orders of magnitude shorter than fatigue lifetimes for the stresses >125 MPa in air (100 MPa in steam). In air, introducing a hold period of up to 100-s duration into a fatigue cycle has no effect on cyclic life for the maximum stress levels 6125 MPa. Run￾out was achieved in all 125 MPa cyclic tests with hold time. The run-out specimens retained 100% of tensile strength. For the stress levels P154 MPa, superposition of a 10-s hold causes a significant reduction in fatigue life. Increas￾ing the hold period to 100 s further reduces fatigue life by an order of magnitude. In steam for the fatigue stress levels P100 MPa, a superposition of a hold time of at least 10-s duration onto a fatigue cycle dramatically degrades cyclic life, reducing it to the much shorter creep life at a given applied stress. Increasing the hold period does not cause further degradation in cyclic life. In air, lowest strains are accumulated in fatigue and highest strains, in creep. Strains accumulated in fatigue tests with hold fall in the intermedi￾ate range. Strain accumulations increase with increasing hold time. In steam, evolution of maximum strain in fati￾gue with hold is akin to that in creep. For a given maxi￾mum stress, strains accumulated in cyclic tests with hold are close to those accumulated in creep and significantly higher than those accumulated in fatigue. Steady-state rate of strain accumulation was reached in all tests. Presence of steam considerably accelerates strain accumulation under both cyclic and static loadings. In air, the rate of strain accumulation increases with hold time at maximum stress. In steam, strain rates in cyclic tests with hold of any dura￾tion are close to those obtained in creep and significantly higher than those observed in fatigue. 4.2. Cyclic–static and static–cyclic block loadings Effects of prior fatigue on creep performance of the N720/A composite were studied in cyclic–static block load￾ing tests at 1200 C for maximum stress levels of 125 and 154 MPa in laboratory air, and for maximum stress levels of 100 and 125 MPa in steam. The fatigue block contained 105 cycles. In air, prior fatigue dramatically improves creep resistance of the CMC, increasing creep life by one to two orders of magnitude. Creep strain rates for the pre-fatigued specimens are at least an order of magnitude lower than those for the as-processed material. For the pre-fatigued composite the run-out stress was 125 MPa, while the as￾processed composite did not achieve a run-out. The pre￾fatigued specimen that achieved a run-out retained over 100% of its tensile strength and 76% of its modulus. In steam, prior fatigue reduced creep resistance of the composite. Effects of prior creep on fatigue durability of the N720/ A composite were studied in static-cyclic block loading tests at 1200 C for the maximum stresses of 125 and 100 MPa in air and in steam, respectively. In air the creep period was 2 h and in steam, 0.75 h. In air, prior creep did not affect fatigue life up to 105 cycles, the fatigue run-out 1436 J.M. Mehrman et al. / Composites Science and Technology 67 (2007) 1425–1438