Availableonlineatwww.sciencedirect.com e Science Direct COMPOSITES CIENCE AND TECHNOLOGY ELSEVIER Composites Science and Technology 67(2007)1425-1438 www.elsevier.com/locate/compscitech Influence of hold times on the elevated-temperature fatigue behavior of an oxide-oxide ceramic composite in air and in steam environment JM.Ehrman , M.B. Ruggles-Wrenn , S.S.Baek b Department of Aeronautics and Astronautics, Air Force Institute of Technology, Wright-Patterson Air Force Base, OH 45433-7765, USA Agency for Defense Development, Daejeon, Republic of Korea Received 2 June 2006: received in revised form 25 August 2006: accepted 7 September 2006 Available online 15 November 2006 Abstract The effect of hold times at maximum stress on fatigue behavior of an oxide-oxide ceramic composite was investigated at 1200Cin laboratory air and in steam environments. The composite consists of a porous alumina matrix reinforced with woven mullite/alumina (NextelM720)fibers, has no interface between the fiber and matrix, and relies on the porous matrix for faw tolerance Tension-tension fatigue tests with a ratio R (minimum to maximum stress)of 0.05, and hold times of 10 and 100 s were performed for fatigue stresses of 125 and 154 MPa in laboratory air, and for fatigue stresses of 100 and 125 MPa in steam environment. Block loading tests incorporating periods of cyclic and sustained loading were carried out to assess the effects of loading history on material behavior and environmental durability. In laboratory air, lives produced in fatigue tests with hold times exceeded those produced in creep but were shorter than those obtained in fatigue. Prior fatigue resulted in an order of magnitude improvement in creep life. Prior creep had no effect on subsequent fatigue life. Presence of steam significantly degraded the material performance In steam, lives produced in fatigue tests with hold times were close to those obtained in creep. Prior fatigue reduced the creep resistance, and prior creep degraded the subsequent fatigue life Composite microstructure, as well as damage and failure mechanisms were investigated. a qualitative spectral analysis showed evidence of silicon species migration from fiber to matrix, especially in steam. o 2006 Elsevier Ltd. All rights reserved Keywords: A Ceramic-matrix composites: A Oxides; B. Fatigue: B Creep: B Mechanical properties 1. Introduction tions. Additionally, the lower densities of CMCs and their ratures, together with a reduced need for Advances in aerospace propulsion technologies have cooling air, allow for improved high-temperature perfor raised the demand for structural materials that have supe- mance when compared to conventional nickel-based super rior long-term mechanical properties and retained proper- alloys [l]. Ceramic-matrix composites are being evaluated les under high temperature, high pressure and various for use in aerospace turbine engines and are likely to be environmental conditions such as moisture. Ceramic- incorporated in combustion chambers and nozzle exten matrix composites(CMCs ), capable of maintaining excel- sions of the advanced rocket propulsion systems [2, 3]. lent strength and fracture toughness at high temperatures Because these applications require exposure to oxidizing are prime candidate materials for such aerospace applica- environments, the thermodynamic stability and oxidation resistance of CMCs are vital issues The views expressed are those of the authors and do not reflect the Non-oxide fiber/non-oxide matrix composites generally ion of the United States Air Force, Department of show poor oxidation resistance [4, 5]. The degradation Defense or the us government Corresponding author. Tel. +1 937 255 3636x4641; fax: +1 937 656 involves oxidation of fibers and fiber coatings, and is typi 4032 cally accelerated by the presence of moisture [6-8]. Using a E-mail address a.ruggles-wrenn(@afit. edu (M.B. Ruggles. non-oxide fiber/oxide matrix or oxide fiber/non-oxide natrix composites generally does not substantially improve 02663538/S. see front matter 2006 Elsevier Ltd. All rights reserved doi:10.1016j.compscitech.2006.09.005
Influence of hold times on the elevated-temperature fatigue behavior of an oxide–oxide ceramic composite in air and in steam environment q J.M. Mehrman a , M.B. Ruggles-Wrenn a,*, S.S. Baek b a Department of Aeronautics and Astronautics, Air Force Institute of Technology, Wright-Patterson Air Force Base, OH 45433-7765, USA b Agency for Defense Development, Daejeon, Republic of Korea Received 2 June 2006; received in revised form 25 August 2006; accepted 7 September 2006 Available online 15 November 2006 Abstract The effect of hold times at maximum stress on fatigue behavior of an oxide–oxide ceramic composite was investigated at 1200 C in laboratory air and in steam environments. The composite consists of a porous alumina matrix reinforced with woven mullite/alumina (NextelTM720) fibers, has no interface between the fiber and matrix, and relies on the porous matrix for flaw tolerance. Tension–tension fatigue tests with a ratio R (minimum to maximum stress) of 0.05, and hold times of 10 and 100 s were performed for fatigue stresses of 125 and 154 MPa in laboratory air, and for fatigue stresses of 100 and 125 MPa in steam environment. Block loading tests incorporating periods of cyclic and sustained loading were carried out to assess the effects of loading history on material behavior and environmental durability. In laboratory air, lives produced in fatigue tests with hold times exceeded those produced in creep but were shorter than those obtained in fatigue. Prior fatigue resulted in an order of magnitude improvement in creep life. Prior creep had no effect on subsequent fatigue life. Presence of steam significantly degraded the material performance. In steam, lives produced in fatigue tests with hold times were close to those obtained in creep. Prior fatigue reduced the creep resistance, and prior creep degraded the subsequent fatigue life. Composite microstructure, as well as damage and failure mechanisms were investigated. A qualitative spectral analysis showed evidence of silicon species migration from fiber to matrix, especially in steam. 2006 Elsevier Ltd. All rights reserved. Keywords: A. Ceramic–matrix composites; A. Oxides; B. Fatigue; B. Creep; B. Mechanical properties 1. Introduction Advances in aerospace propulsion technologies have raised the demand for structural materials that have superior long-term mechanical properties and retained properties under high temperature, high pressure and various environmental conditions such as moisture. Ceramic– matrix composites (CMCs), capable of maintaining excellent strength and fracture toughness at high temperatures are prime candidate materials for such aerospace applications. Additionally, the lower densities of CMCs and their higher use temperatures, together with a reduced need for cooling air, allow for improved high-temperature performance when compared to conventional nickel-based superalloys [1]. Ceramic–matrix composites are being evaluated for use in aerospace turbine engines and are likely to be incorporated in combustion chambers and nozzle extensions of the advanced rocket propulsion systems [2,3]. Because these applications require exposure to oxidizing environments, the thermodynamic stability and oxidation resistance of CMCs are vital issues. Non-oxide fiber/non-oxide matrix composites generally show poor oxidation resistance [4,5]. The degradation involves oxidation of fibers and fiber coatings, and is typically accelerated by the presence of moisture [6–8]. Using a non-oxide fiber/oxide matrix or oxide fiber/non-oxide matrix composites generally does not substantially improve 0266-3538/$ - see front matter 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.compscitech.2006.09.005 q The views expressed are those of the authors and do not reflect the official policy or position of the United States Air Force, Department of Defense or the US Government. * Corresponding author. Tel.: +1 937 255 3636x4641; fax: +1 937 656 4032. E-mail address: marina.ruggles-wrenn@afit.edu (M.B. RugglesWrenn). www.elsevier.com/locate/compscitech Composites Science and Technology 67 (2007) 1425–1438 COMPOSITES SCIENCE AND TECHNOLOGY
J M. Ehrman et al Composites Science and Technology 67(2007)1425-1438 the high temperature oxidation resistance [9]. The need for San Diego, CA)consisting of a porous alumina matrix environmentally stable composites motivated the develop- reinforced with NextelM720 fibers. There is no fiber coat ment of CMCs based on environmentally stable oxide con- ing. The damage tolerance of N720/A is enabled by the stituents [10-16]. porous matrix. The composite was supplied in a form of o The main advantages of CMCs over monolithic ceram- 2.8-mm thick plates comprised of 12 0/90 woven layers, are their superior toughness, tolerance to the presence with a density of 2.77 g/cm,, a fiber volume of 45% of defects, and non-catastrophic mode of failure. It is and matrix porosity of w24%. The fiber fabric was infil- widely accepted that in order to avoid brittle fracture trated with the matrix in a sol-gel process. The laminate behavior in CMCs and improve the damage tolerance, a was dried with a"vacuum bag"technique under low pres- weak fiber/matrix interface is needed, which serves to sure and low temperature then pressureless sintered [28]. deflect matrix cracks and to allow subsequent fiber pull- Representative micrographs of the untested material are out [17-20]. It has recently been demonstrated that a shown in Fig. 1. Fig. la shows 00 and 90 fiber tows as well matrix with finely distributed porosity can provide a simi- as numerous matrix cracks In the case of the as-processed lar crack-deflecting behavior [21]. This microstructural material, most are shrinkage cracks formed during process design philosophy implicitly accepts the strong fiber /matrix ing rather than matrix cracks generated during loading interface. The concept has been successfully demonstrated Porous nature of the matrix is seen in Fig. 1b for oxide-oxide composites [10, 13, 16, 22-24]. Resulting A servocontrolled MTS mechanical testing machine oxide/oxide CMCs exhibit damage tolerance combined equipped with hydraulic water-cooled collet grips, a com- ith inherent oxidation resistance. An extensive review of pact two-zone resistance-heated furnace, and two tempera- the mechanisms and mechanical properties of porous- ture controllers were used in all tests. An MTS TestStar II natrix CMCs is given in [25] digital controller was employed for input signal generation Porous-matrix oxide/oxide CMCs exhibit several behav- and data acquisition. Strain measurement was accom- ior trends that are distinctly different from those exhibited plished with an MTS high-temperature air-cooled uniaxial by traditional CMCs with a fiber-matrix interface. For extensometer. For elevated temperature testing, thermo these materials, fatigue is significantly more damaging than couples were bonded to test specimens to calibrate the fur- creep Zawada et al. [26]examined the mechanical behavior nace on a periodic basis. The furnace controller(using a of a porous-matrix Nextel610/Alumino-silicate composite non-contacting thermocouple exposed to the ambient envi- at 1000C and reported that the CMc exhibited excellent ronment near the test specimen)was adjusted to determine fatigue performance, but low creep resistance. Ruggles- the power setting needed to achieve the desired tempera- Wrenn et al. [27] investigated fatigue and creep behaviors ture of the test specimen. The determined power setting of N720/A composite at 1200C. The results revealed was then used in actual tests. The power setting for testing excellent fatigue resistance in laboratory air, the material in steam environment was determined by placing the spec exhibited high fatigue limit, long fatigue life and near imen instrumented with thermocouples in steam and 100% strength retention. However, N720/A produced short repeating the furnace calibration procedure. Thermocou- creep lives, revealing low creep resistance at 1200oC. Pres- ples were not bonded to the test specimens after the furnace ence of steam significantly degraded the fatigue as well as was calibrated. Tests in steam employed an alumina sus- the creep performance of the CMC ceptor(tube with end caps), which fits inside the furnace. Because creep was shown to be considerably more dam- The specimen gage section is located inside the susceptor aging than cyclic loading to porous-matrix oxide-oxide with the ends of the specimen passing through slots in CMCs [26, 27), mechanical performance under combined the susceptor. Steam is introduced into the susceptor static-cyclic loading in high-temperature oxidizing environ-(through a feeding tube) in a continuous stream with a ments remains among the key issues that must be addressed slightly positive pressure, expelling the dry air and creating before using these materials in aerospace applications. The a 100% steam environment inside the susceptor objective of this study is to investigate the influence of hold All tests were performed at 1200C. Each test specimen imes at maximum stress on fatigue performance of Nex- was heated to 1200C in 25 min, and held at 1200C for [T720/Alumina, a porous-matrix oxide-oxide CFCC, in additional 15 min prior to testing. Dog bone shaped speci laboratory air and in steam environments. In addition, this mens of 152-mm total length with a 10-mm-wide gage sec- effort aims to assess the effects of prior loading history on tion were used in all tests. Tensile tests were performed in mechanical behavior and durability of N720/. The com- displacement control with a constant rate of 0.05 mm/s in posite microstructure, as well as damage and failure mech- laboratory air. All creep-fatigue interaction tests were per anisms are also discussed formed in load control and employed the ratio R(mini- mum to maximum stress of 0.05, and the loading/ unloading frequency of 1 Hz. Cyclic test with hold time 2. Experimental procedure consists of a standard fatigue cycle with a hold time of either 10 or 100 s superimposed at the maximum stress The material studied was NextelTM720/Alumina(N720/ The run-out is set at 100 h, which is consistent with the ser A), an oxide-oxide CMC (manufactured by COI Ceramics, vice life expected in aerospace applications at that temper
the high temperature oxidation resistance [9]. The need for environmentally stable composites motivated the development of CMCs based on environmentally stable oxide constituents [10–16]. The main advantages of CMCs over monolithic ceramics are their superior toughness, tolerance to the presence of defects, and non-catastrophic mode of failure. It is widely accepted that in order to avoid brittle fracture behavior in CMCs and improve the damage tolerance, a weak fiber/matrix interface is needed, which serves to deflect matrix cracks and to allow subsequent fiber pullout [17–20]. It has recently been demonstrated that a matrix with finely distributed porosity can provide a similar crack-deflecting behavior [21]. This microstructural design philosophy implicitly accepts the strong fiber/matrix interface. The concept has been successfully demonstrated for oxide–oxide composites [10,13,16,22–24]. Resulting oxide/oxide CMCs exhibit damage tolerance combined with inherent oxidation resistance. An extensive review of the mechanisms and mechanical properties of porousmatrix CMCs is given in [25]. Porous-matrix oxide/oxide CMCs exhibit several behavior trends that are distinctly different from those exhibited by traditional CMCs with a fiber–matrix interface. For these materials, fatigue is significantly more damaging than creep. Zawada et al. [26] examined the mechanical behavior of a porous-matrix Nextel610/Alumino-silicate composite at 1000 C and reported that the CMC exhibited excellent fatigue performance, but low creep resistance. RugglesWrenn et al. [27] investigated fatigue and creep behaviors of N720/A composite at 1200 C. The results revealed excellent fatigue resistance in laboratory air, the material exhibited high fatigue limit, long fatigue life and near 100% strength retention. However, N720/A produced short creep lives, revealing low creep resistance at 1200 C. Presence of steam significantly degraded the fatigue as well as the creep performance of the CMC. Because creep was shown to be considerably more damaging than cyclic loading to porous-matrix oxide–oxide CMCs [26,27], mechanical performance under combined static-cyclic loading in high-temperature oxidizing environments remains among the key issues that must be addressed before using these materials in aerospace applications. The objective of this study is to investigate the influence of hold times at maximum stress on fatigue performance of NextelTM720/Alumina, a porous-matrix oxide–oxide CFCC, in laboratory air and in steam environments. In addition, this effort aims to assess the effects of prior loading history on mechanical behavior and durability of N720/A. The composite microstructure, as well as damage and failure mechanisms are also discussed. 2. Experimental procedure The material studied was NextelTM720/Alumina (N720/ A), an oxide–oxide CMC (manufactured by COI Ceramics, San Diego, CA) consisting of a porous alumina matrix reinforced with NextelTM720 fibers. There is no fiber coating. The damage tolerance of N720/A is enabled by the porous matrix. The composite was supplied in a form of 2.8-mm thick plates comprised of 12 0/90 woven layers, with a density of 2.77 g/cm3 , a fiber volume of 45%, and matrix porosity of 24%. The fiber fabric was infiltrated with the matrix in a sol-gel process. The laminate was dried with a ‘‘vacuum bag’’ technique under low pressure and low temperature, then pressureless sintered [28]. Representative micrographs of the untested material are shown in Fig. 1. Fig. 1a shows 0 and 90 fiber tows as well as numerous matrix cracks. In the case of the as-processed material, most are shrinkage cracks formed during processing rather than matrix cracks generated during loading. Porous nature of the matrix is seen in Fig. 1b. A servocontrolled MTS mechanical testing machine equipped with hydraulic water-cooled collet grips, a compact two-zone resistance-heated furnace, and two temperature controllers were used in all tests. An MTS TestStar II digital controller was employed for input signal generation and data acquisition. Strain measurement was accomplished with an MTS high-temperature air-cooled uniaxial extensometer. For elevated temperature testing, thermocouples were bonded to test specimens to calibrate the furnace on a periodic basis. The furnace controller (using a non-contacting thermocouple exposed to the ambient environment near the test specimen) was adjusted to determine the power setting needed to achieve the desired temperature of the test specimen. The determined power setting was then used in actual tests. The power setting for testing in steam environment was determined by placing the specimen instrumented with thermocouples in steam and repeating the furnace calibration procedure. Thermocouples were not bonded to the test specimens after the furnace was calibrated. Tests in steam employed an alumina susceptor (tube with end caps), which fits inside the furnace. The specimen gage section is located inside the susceptor, with the ends of the specimen passing through slots in the susceptor. Steam is introduced into the susceptor (through a feeding tube) in a continuous stream with a slightly positive pressure, expelling the dry air and creating a 100% steam environment inside the susceptor. All tests were performed at 1200 C. Each test specimen was heated to 1200 C in 25 min, and held at 1200 C for additional 15 min prior to testing. Dog bone shaped specimens of 152-mm total length with a 10-mm-wide gage section were used in all tests. Tensile tests were performed in displacement control with a constant rate of 0.05 mm/s in laboratory air. All creep–fatigue interaction tests were performed in load control and employed the ratio R (minimum to maximum stress) of 0.05, and the loading/ unloading frequency of 1 Hz. Cyclic test with hold time consists of a standard fatigue cycle with a hold time of either 10 or 100 s superimposed at the maximum stress. The run-out is set at 100 h, which is consistent with the service life expected in aerospace applications at that temper- 1426 J.M. Mehrman et al. / Composites Science and Technology 67 (2007) 1425–1438
J.M. Ehrman et aL/Ce tes Science and Technology 67(2007)1425-1438 r Fig. 1. As-received material:(a)overview, optical microscope and (b) porous nature of the matrix is evident(SEM ature [29]. Cyclic-static block loading test block of 10 fatigue cycles followed by static le loadin ses a and 100 s were selected to simulate realistic conditions for with turbine ts[29,31 the same maximum stress. The run-out is defined as Results are summarized in Table l where test environ- 100 h at creep stress Static-cyclic block loading test con- ment, test type and maximum stress are shown together sists of static loading(2 h in air and 0.75 h in steam) fol- with time to failure. Results are also presented in Figs. 2a lowed by cyclic loading with the same maximum stress. and b as maximum stress vs time to failure curves for air In this case the run-out is defined as survival of 105 fatigue and steam, respectively. Fatigue and creep-rupture results cycles. In each test, stress-strain data were recorded during cyclic loading as well as during creep periods. Thus stifness degradation and strain accumulation with fatigue cycles Table 1 achieved a run-out were subjected to tensile test to failure composite at /on its of creep-fatigue interaction tests for the N720/A and/or time could be examined. All specimens that Summary of res and steam environments at 1200C to determine the retained strength and stiffness. Test type Max stress(MPa Fracture surfaces of failed specimens were examined Laboratory air ng SEM(FEI Quanta 200 HV) as well as an optical Fatigue, 10 s hold The SeM 1464 128a Energy-dispersive X-ray spectroscopy (EDS) analysis was Fatigue. 100s hold 112 performed using an EDAX Genesis 4000 EDS system 54 17.6 Fatigue. 100s hold 154 2.75 3. Results and discussion Fatigue 125 40.7° Fat 154 C 125 4.25 3.. Monotonic tension 154 Fatigue-creep 125 Tensile results obtained at 1200C were consistent with Fatigue-creep those reported earlier [27, 30]. The ultimate tensile strength Steam environment (UTS)was 190 MPa, elastic modulus, 76 GPa, and failure Fatigue, 10s hold 4.63 strain, 0.38%. It is worthy of note that in all tests reported Fatigue, 10s hold herein,the failure occurred within the gage section of the Fatigue, 100 s hold Fatigue. 100s hold Fatigue. 10s hold 0005 1.12 Fatigue. 100s hold 3. 2. Fatigue with hold times F 30.0 F Fatigue with hold times at maximum stress were CI 2.49 conducted at 1200°C and in steam. For each envi- creep b 1.25 0.24 ronment, fatigue stress levels were selected according to Fatigue-creep 100 0.61 the results of the previous study [27] to be at or below the fatigue limit, yet above the creep run-out stress. Hence, comparison Results of fatigue and creep tests from prior work [27] are included fo stress levels of 125 and 154 MPa were used in air and stress Run-out levels of 100 and 125 MPa. in steam the hold times of 10 From Ruggles-Wrenn et al. [27]
ature [29]. Cyclic–static block loading test comprises a block of 105 fatigue cycles followed by static loading with the same maximum stress. The run-out is defined as 100 h at creep stress. Static-cyclic block loading test consists of static loading (2 h in air and 0.75 h in steam) followed by cyclic loading with the same maximum stress. In this case the run-out is defined as survival of 105 fatigue cycles. In each test, stress-strain data were recorded during cyclic loading as well as during creep periods. Thus stiffness degradation and strain accumulation with fatigue cycles and/or time could be examined. All specimens that achieved a run-out were subjected to tensile test to failure at 1200 C to determine the retained strength and stiffness. Fracture surfaces of failed specimens were examined using SEM (FEI Quanta 200 HV) as well as an optical microscope. The SEM specimens were carbon coated. Energy-dispersive X-ray spectroscopy (EDS) analysis was performed using an EDAX Genesis 4000 EDS system. 3. Results and discussion 3.1. Monotonic tension Tensile results obtained at 1200 C were consistent with those reported earlier [27,30]. The ultimate tensile strength (UTS) was 190 MPa, elastic modulus, 76 GPa, and failure strain, 0.38%. It is worthy of note that in all tests reported herein, the failure occurred within the gage section of the extensometer. 3.2. Fatigue with hold times Fatigue tests with hold times at maximum stress were conducted at 1200 C in air and in steam. For each environment, fatigue stress levels were selected according to the results of the previous study [27] to be at or below the fatigue limit, yet above the creep run-out stress. Hence, stress levels of 125 and 154 MPa were used in air, and stress levels of 100 and 125 MPa, in steam. The hold times of 10 and 100 s were selected to simulate realistic conditions for turbine engine exhaust components [29,31]. Results are summarized in Table 1, where test environment, test type and maximum stress are shown together with time to failure. Results are also presented in Figs. 2a and b as maximum stress vs time to failure curves for air and steam, respectively. Fatigue and creep-rupture results Fig. 1. As-received material: (a) overview, optical microscope and (b) porous nature of the matrix is evident (SEM). Table 1 Summary of results of creep–fatigue interaction tests for the N720/A composite at 1200 C in laboratory air and steam environments Test type Max stress (MPa) Time to failure (h) Laboratory air Fatigue, 10 s hold 125 103a Fatigue, 10 s hold 125 146a Fatigue, 100 s hold 125 128a Fatigue, 100 s hold 125 112a Fatigue, 10 s hold 154 17.6 Fatigue, 100 s hold 154 2.75 Fatigue b 125 40.7a Fatigue b 154 46.5a Creep b 125 4.25 Creep b 154 0.27 Fatigue–creep 125 101a Fatigue–creep 154 1.68 Steam environment Fatigue, 10 s hold 100 4.63 Fatigue, 10 s hold 100 2.36 Fatigue, 100 s hold 100 1.35 Fatigue, 100 s hold 100 1.12 Fatigue, 10 s hold 125 0.22 Fatigue, 100 s hold 125 0.31 Fatigue b 100 30.0a Fatigue b 125 46.2a Creep b 100 2.49 Creep 100 1.25 Creep b 125 0.24 Fatigue–creep 100 0.61 Results of fatigue and creep tests from prior work [27] are included for comparison. a Run-out. b From Ruggles-Wrenn et al. [27]. J.M. Mehrman et al. / Composites Science and Technology 67 (2007) 1425–1438 1427
J M. Ehrman et al Composites Science and Technology 67(2007)1425-1438 from prior study [27] are included for comparison. Data in Table I and Fig. 2a demonstrate that in air static(creep) T=1200C. Air loading is significantly more damaging than cyclic loading For a given stress level, specimens tested in creep exhibit a much shorter lives than those tested in fatigue or fatigue a with hold time. Introduction of a hold period into the fati- gue cycle had no noticeable effect on time to failure for the g stress level of 125 MPa. All 125 MPa fatigue tests, regard less of the hold time, achieved a 100-h run-out. Apparently at this stress level, a 100-s hold time is not long enough to vith 100s Hold allow appreciable creep damage to develop and signifi uggles-Wrenn et al, 2006 cantly affect cyclic life. However, for the stress level of the specimen tested in fatigue achieved a run-out of 10sa07 154 MPa, superposition of even a shorter 10-s hold period 1000 onto the fatigue cycle noticeably reduced cyclic life. while cycles(28 h at 1 Hz), specimen subjected to fatigue with T= 1200 C Steam 10-s hold failed after 17.6 h. At 154 MPa, a 10-s hold per- showed that at 1200C creep strain rate of N720/A was 4@ 200 lod is sufficient to initiate creep damage. Prior study [27] UTS strongly dependent on the applied stress; increase in creep stress from 125 to 154 MPa caused an order of magnitude increase in creep rate Higher creep rate is likely responsible 100 A口 for acceleration of damage growth seen in the 154 fatigue 9 口 Fatigue, Ruggles-Wrenn et al,2001△ test with hold. Increasing the hold time at 154 MPa by an order of magnitude resulted in an order of magnitude ■ Fatigue with100 s Hold reduction in life. The specimen subjected to fatigue with 100-s hold survived only 2.73 h. It appears that creep dam 1000 age generated during the 100-s hold period of the first cycle b Time(h) was significant and continued to accumulate more rapidly as the cycling progressed. Finally in the case of the creep at 120 Fig. 2. Maximum stress vs time to failure for N720/A ceramic composite test,which represents the most damaging loading type, life shoa -C: (a)in air and (b)in steam Creep and fatigue data ([27] are also is reduced by another order of magnitude to 0. 27h. The presence of steam noticeably degrades fati well as creep performance of the CMC. Results also reveal loading would govern damage development and that in steam static loading remains considerably more creep would have little effect damaging than cyclic loading. At 154 MPa, the failure time Prior investigation [27] reported substantial strain accu- in fatigue test was 3.27 h, two orders of magnitude higher mulation in creep as well as ratcheting under cyclic loading than the failure time of 0.027 h obtained in creep [27]. at 1200C. Present results permit further assessment of the Superposition of a hold time onto a fatigue cycle dramati- effect of loading type on the rate of strain accumulation. cally degrades cyclic life in steam. At 100 MPa, fatigue test Maximum strain(i.e. strain at maximum stress)as a func achieved a run-out, surviving at least 30 h, while two fati- tion of time for cyclic tests conducted in air is presented in gue tests with 10-s hold failed after 2.36 and 4.63 h Intro- Fig 3, where creep and fatigue test results from [27]are ducing a 10-s hold reduces the cyclic life by an order of also included. Regardless of the maximum stress, strain magnitude bringing it down to the lifetimes obtained in accumulations are lowest in fatigue and highest in creep, creep. Note that fatigue tests with 100-s hold produced sim- while strains accumulated in fatigue with hold time fall in ilar lifetimes(1. 12 and 1.35 h). Similar observations can be the intermediate range. As expected increasing the hold made for the 125 MPa tests. While a 125 MPa fatigue test time results in larger strain accumulations. Maximum achieved a run-out surviving at least 46.2 h, fatigue tests strain vs time for tests conducted in steam is shown in with 10-s and 100-s hold failed after only 0.23 and 0.33 h, Fig. 4. In steam, evolution of maximum strain in cyclic respectively. These failure times are close to that produced tests with hold time is akin to that observed in creep at in creep. Apparently in steam, for stress levels >100 MPa, the same applied stress. This trend becomes more pro a 10-s hold is sufficiently long to activate the damage pro- nounced at higher stress levels. Conversely, fatigue tests cess associated with creep, which proceeds to dominate the produced little strain accumulation. These observations damage development and govern the cyclic life. Once the indicate that in air as well as in steam, the prevailing dam- said damage process sets in, increasing the hold time has age mechanisms operate under sustained loading. Of no appreciable effect on cyclic life. It is recognized that importance are the rates of strain accumulation, which some lower value of the hold time may exist below which reflect damage development under load and serve as indica
from prior study [27] are included for comparison. Data in Table 1 and Fig. 2a demonstrate that in air static (creep) loading is significantly more damaging than cyclic loading. For a given stress level, specimens tested in creep exhibit much shorter lives than those tested in fatigue or fatigue with hold time. Introduction of a hold period into the fatigue cycle had no noticeable effect on time to failure for the stress level of 125 MPa. All 125 MPa fatigue tests, regardless of the hold time, achieved a 100-h run-out. Apparently at this stress level, a 100-s hold time is not long enough to allow appreciable creep damage to develop and signifi- cantly affect cyclic life. However, for the stress level of 154 MPa, superposition of even a shorter 10-s hold period onto the fatigue cycle noticeably reduced cyclic life. While the specimen tested in fatigue achieved a run-out of 105 cycles (28 h at 1 Hz), specimen subjected to fatigue with 10-s hold failed after 17.6 h. At 154 MPa, a 10-s hold period is sufficient to initiate creep damage. Prior study [27] showed that at 1200 C creep strain rate of N720/A was strongly dependent on the applied stress; increase in creep stress from 125 to 154 MPa caused an order of magnitude increase in creep rate. Higher creep rate is likely responsible for acceleration of damage growth seen in the 154 fatigue test with hold. Increasing the hold time at 154 MPa by an order of magnitude resulted in an order of magnitude reduction in life. The specimen subjected to fatigue with 100-s hold survived only 2.73 h. It appears that creep damage generated during the 100-s hold period of the first cycle was significant and continued to accumulate more rapidly as the cycling progressed. Finally in the case of the creep test, which represents the most damaging loading type, life is reduced by another order of magnitude to 0.27 h. The presence of steam noticeably degrades fatigue as well as creep performance of the CMC. Results also reveal that in steam static loading remains considerably more damaging than cyclic loading. At 154 MPa, the failure time in fatigue test was 3.27 h, two orders of magnitude higher than the failure time of 0.027 h obtained in creep [27]. Superposition of a hold time onto a fatigue cycle dramatically degrades cyclic life in steam. At 100 MPa, fatigue test achieved a run-out, surviving at least 30 h, while two fatigue tests with 10-s hold failed after 2.36 and 4.63 h. Introducing a 10-s hold reduces the cyclic life by an order of magnitude bringing it down to the lifetimes obtained in creep. Note that fatigue tests with 100-s hold produced similar lifetimes (1.12 and 1.35 h). Similar observations can be made for the 125 MPa tests. While a 125 MPa fatigue test achieved a run-out surviving at least 46.2 h, fatigue tests with 10-s and 100-s hold failed after only 0.23 and 0.33 h, respectively. These failure times are close to that produced in creep. Apparently in steam, for stress levels P100 MPa, a 10-s hold is sufficiently long to activate the damage process associated with creep, which proceeds to dominate the damage development and govern the cyclic life. Once the said damage process sets in, increasing the hold time has no appreciable effect on cyclic life. It is recognized that some lower value of the hold time may exist below which cyclic loading would govern damage development and creep would have little effect. Prior investigation [27] reported substantial strain accumulation in creep as well as ratcheting under cyclic loading at 1200 C. Present results permit further assessment of the effect of loading type on the rate of strain accumulation. Maximum strain (i. e. strain at maximum stress) as a function of time for cyclic tests conducted in air is presented in Fig. 3, where creep and fatigue test results from [27] are also included. Regardless of the maximum stress, strain accumulations are lowest in fatigue and highest in creep, while strains accumulated in fatigue with hold time fall in the intermediate range. As expected increasing the hold time results in larger strain accumulations. Maximum strain vs time for tests conducted in steam is shown in Fig. 4. In steam, evolution of maximum strain in cyclic tests with hold time is akin to that observed in creep at the same applied stress. This trend becomes more pronounced at higher stress levels. Conversely, fatigue tests produced little strain accumulation. These observations indicate that in air as well as in steam, the prevailing damage mechanisms operate under sustained loading. Of importance are the rates of strain accumulation, which reflect damage development under load and serve as indica- 0 50 100 150 200 250 0.1 1 10 100 1000 Time (h) Max Stress (MPa) Fatigue, Ruggles-Wrenn et al, 2006 Fatigue with 10 s Hold Fatigue with 100 s Hold Creep, Ruggles-Wrenn et al, 2006 UTS T = 1200˚C, Air 0 50 100 150 200 250 0.01 0.1 1 10 100 1000 Time (h) Max Stress (MPa) Fatigue, Ruggles-Wrenn et al, 2006 Fatigue with 10 s Hold Fatigue with 100 s Hold Creep, Ruggles-Wrenn et al, 2006 UTS T = 1200˚C, Steam Fig. 2. Maximum stress vs time to failure for N720/A ceramic composite at 1200 C: (a) in air and (b) in steam. Creep and fatigue data [27] are also shown. 1428 J.M. Mehrman et al. / Composites Science and Technology 67 (2007) 1425–1438
J.M. Ehrman et al. Composites Science and Technology 67(2007)1425-1438 1429 T=1200℃c,Air T= 1200.C. Steam Max stress 125 MPa Max Stress= 100 MPa uggles-Wrenn, 2006 100s Hold Fatigue with 100 s Hold 0.50 0.00 Time(h) T=1200℃c,Ai T=1200℃c, Stean Max Stress= 125 MPa 0.75 Creep, Ruggles-Wrenn 2006 1.5 old 0.50 Fatigue with 10 s Hold 100 s Hold uggles-Wrenn 2006 Fatigue, Ruggles-Wrenn 2006 0.00 0.0 Time(h) 0.00 0.75 Time(h) Fig 3. Maximum strain as a function of time at 1200C in laboratory ai for the maximum stress of:(a)125 MPa and(b)154 MPa. Creep and ig. 4. Maximum strain as a function of time at 1200oC in steam Fi fatigue data [27] are also shown environment for the maximum stress of: (a)100 MPa and(b)125 MPa. Creep and fatigue data [27] are also shown. tors of how damaging the particular type of loading is Note that the steady-state strain rate was reached in all 1.0E-02 tests shown in Figs. 3 and 4. The steady-state strain rates 口 Steam, Fatigue as functions of maximum stress are presented in Fig. 5 1.0E-03 △ Steam, e 10s Hold In air. the rate of strain accumulation increases with unin ◆ Stean, e 100s Hold terrupted time spent at maximum stress. Introducing a 10-s hold into a fatigue cycle causes an increase in strain rate Increasing the hold time from 10 to 100 s results in a fur ther(nearly an order of magnitude) increase in strain rate 10E-06 In creep tests(a further increase in hold time) strain rate 10E-07 increases by yet another order of magnitude. These obser ations are consistent with results reported by Zawada 1.0E-08 et al. for N610/AS composite T=1200 1.0E-09 Presence of steam generally accelerates strain accumula ion under both cyclic and sustained loadings. Strain rates Max Stress(MPa tude higher than those obtained in air for a given fatigue g. s sd an rate as a intron en mumum appied stress at times of increasing duration are introduced into a fatigue cycle. In steam, strain rates produced in fatigue with hold ate, resulting in higher rates of strain accumulation. The time of any duration are close to those obtained in creep presence of steam accelerates damage initiation and and significantly higher than those obtained in fatigue. growth The damaging nature of sustained loading is evident. H Retained strength and stiffness of the specimens, which periods allow time-dependent damage mechanisms to initi- achieved run-out in fatigue tests with hold time, are sum-
tors of how damaging the particular type of loading is. Note that the steady-state strain rate was reached in all tests shown in Figs. 3 and 4. The steady-state strain rates as functions of maximum stress are presented in Fig. 5. In air, the rate of strain accumulation increases with uninterrupted time spent at maximum stress. Introducing a 10-s hold into a fatigue cycle causes an increase in strain rate. Increasing the hold time from 10 to 100 s results in a further (nearly an order of magnitude) increase in strain rate. In creep tests (a further increase in hold time) strain rate increases by yet another order of magnitude. These observations are consistent with results reported by Zawada et al. for N610/AS composite [29]. Presence of steam generally accelerates strain accumulation under both cyclic and sustained loadings. Strain rates obtained in fatigue in steam are at least an order of magnitude higher than those obtained in air for a given fatigue stress. In air, the strain rates increase gradually when hold times of increasing duration are introduced into a fatigue cycle. In steam, strain rates produced in fatigue with hold time of any duration are close to those obtained in creep and significantly higher than those obtained in fatigue. The damaging nature of sustained loading is evident. Hold periods allow time-dependent damage mechanisms to initiate, resulting in higher rates of strain accumulation. The presence of steam accelerates damage initiation and growth. Retained strength and stiffness of the specimens, which achieved run-out in fatigue tests with hold time, are sum- 0.00 0.25 0.50 0.75 1.00 1.25 1.50 0 10 20 30 40 50 60 Time (h) Strain (%) T = 1200˚C, Air Max Stress = 125 MPa Creep Ruggles-Wrenn, 2006 Fatigue with 100 s Hold Fatigue with 10 s Hold Fatigue, Ruggles-Wrenn 2006 0.00 0.25 0.50 0.75 1.00 01234 Time (h) Strain (%) Creep, Ruggles-Wrenn 2006 Fatigue with 100 s Hold Fatigue with 10 s Hold T = 1200˚C, Air Max Stress = 154 MPa Fatigue, Ruggles-Wrenn 2006 Fig. 3. Maximum strain as a function of time at 1200 C in laboratory air for the maximum stress of: (a) 125 MPa and (b) 154 MPa. Creep and fatigue data [27] are also shown. 0.0 0.5 1.0 1.5 2.0 2.5 012345 Time (h) Strain (%) T = 1200˚C, Steam Max Stress = 100 MPa Creep Ruggles-Wrenn 2006 Fatigue Ruggles-Wrenn 2006 Fatigue 100 s Hold Fatigue 10 s Hold 0.0 0.5 1.0 1.5 2.0 0.00 0.25 0.50 0.75 1.00 Time (h) Strain (%) T = 1200˚C, Steam Max Stress = 125 MPa Creep Ruggles-Wrenn 2006 Fatigue Ruggles-Wrenn 2006 Fatigue 10 s Hold Fatigue 100 s Hold Fig. 4. Maximum strain as a function of time at 1200 C in steam environment for the maximum stress of: (a) 100 MPa and (b) 125 MPa. Creep and fatigue data [27] are also shown. 1.0E-09 1.0E-08 1.0E-07 1.0E-06 1.0E-05 1.0E-04 1.0E-03 1.0E-02 10 100 1000 Max Stress (MPa) Strain Rate (s-1) Air, Fatigue Steam, Fatigue Air, Fatigue 10 s Hold Steam, Fatigue 10 s Hold Air, Fatigue 100 s Hold Steam, Fatigue 100 s Hold Air, Creep Steam, Creep T = 1200˚C Fig. 5. Strain rate as a function of maximum applied stress at 1200 C. Creep and fatigue data [27] are also shown. J.M. Mehrman et al. / Composites Science and Technology 67 (2007) 1425–1438 1429
J M. Ehrman et al Composites Science and Technology 67(2007)1425-1438 Table 2 Retained properties of the N720/A specimens subjected to creep-fatigue interaction tests with the maximum stress of 125 MPa in laboratory air at 1200C Prior test duration (h) etained strength(MPa) Retained modulus(GPa) Failure strain (% 574 Fatigue, 10 s hold 38 216 Fatigue, 100s hold l12 .18 Cyclic-static l.04 marized in Table 2. All specimens tested r retained Stress-rupture behavior of the pre-fatigued and the as- 00% of their tensile strength. However, stiffness loss of processed specimens tested in air is summarized in Fig. 6 26-35% was observed, which appears to be independent In air prior fatigue has a dramatic effect on the subsequent of the maximum stress or the hold time. Full retention of creep life. For a given applied stress, creep lives of the pre- tensile strength indicates that no damage occurred to the fatigued specimens are one to two orders of magnitude fibers. The residual modulus of 60 GPa is consistent with longer than those of the as-processed specimens. At loading direction, indicating saturation of microcracks in out of 100 h, while the d specimen achieved a creep run- the matrix and no load transfer within the 90 tows. In only 4.25 h At 154 MPa, rupture time for the pre-fatigued team, run-out was not achieved in fatigue tests with hold specimen was 1.68h, an order of magnitude higher than the rupture time of 0. 27 h for the as-processed specimen The pre-fatigued specimen that achieved creep run-out at 3.3. Cyclic-static block loadings 125 MPa was subjected to a tensile test to failure at Previous study [27] revealed a significant difference in time to failure under sustained and cyclic loadings. Fatigue run-out of 10 cycles(x28 h at 1 Hz) was achieved at max- imum stress levels that caused creep failure in less than I h T=1200C. Air Cyclic-static block loading tests were conducted to evalu ate the effects of prior fatigue on creep performance. The number of cycles in the cyclic block was 10, the same as As-Processed. 154 MPa the fatigue run-out condition. The maximum stress levels o were 125 and 154 MPa in air. and 100 MPa in steam Results are included in Table 1. Because all specimens sub jected to cyclic-static tests survived the 10 cycles of the fatigue block, time to failure in Table I is in fact the time to failure obtained in the creep block. The total times to I Hz)to the creep rupture times given in Table/ ycles at failure can be calculated by adding 27.7 h(105 Time(h) T=1200C. Air ooC. stear 200IUTS 1000 Time(h) Time(h) Fig. 6. Creep stress vs time to rupture for Nextel 720/Alumina ceramic Fig. 7. Creep strain vs time for Nextel 720/Alumina ceramic composite composite at 1200C in laboratory air data from Ruggles- Wrenn at 1200C:(a)in laboratory air, (b) in steam environment. Creep data et al. [27) from Ruggles-Wrenn et al. [27]are also shown
marized in Table 2. All specimens tested in air retained 100% of their tensile strength. However, stiffness loss of 26–35% was observed, which appears to be independent of the maximum stress or the hold time. Full retention of tensile strength indicates that no damage occurred to the fibers. The residual modulus of 60 GPa is consistent with the composite modulus being due entirely to fibers in the loading direction, indicating saturation of microcracks in the matrix and no load transfer within the 90 tows. In steam, run-out was not achieved in fatigue tests with hold time. 3.3. Cyclic–static block loadings Previous study [27] revealed a significant difference in time to failure under sustained and cyclic loadings. Fatigue run-out of 105 cycles (28 h at 1 Hz) was achieved at maximum stress levels that caused creep failure in less than 1 h. Cyclic–static block loading tests were conducted to evaluate the effects of prior fatigue on creep performance. The number of cycles in the cyclic block was 105 , the same as the fatigue run-out condition. The maximum stress levels were 125 and 154 MPa in air, and 100 MPa in steam. Results are included in Table 1. Because all specimens subjected to cyclic–static tests survived the 105 cycles of the fatigue block, time to failure in Table 1 is in fact the time to failure obtained in the creep block. The total times to failure can be calculated by adding 27.7 h (105 cycles at 1 Hz) to the creep rupture times given in Table 1. Stress-rupture behavior of the pre-fatigued and the asprocessed specimens tested in air is summarized in Fig. 6. In air prior fatigue has a dramatic effect on the subsequent creep life. For a given applied stress, creep lives of the prefatigued specimens are one to two orders of magnitude longer than those of the as-processed specimens. At 125 MPa, the pre-fatigued specimen achieved a creep runout of 100 h, while the as-processed specimen failed after only 4.25 h. At 154 MPa, rupture time for the pre-fatigued specimen was 1.68 h, an order of magnitude higher than the rupture time of 0.27 h for the as-processed specimen. The pre-fatigued specimen that achieved creep run-out at 125 MPa was subjected to a tensile test to failure at Table 2 Retained properties of the N720/A specimens subjected to creep–fatigue interaction tests with the maximum stress of 125 MPa in laboratory air at 1200 C Prior test type Prior test duration (h) Retained strength (MPa) Retained modulus (GPa) Failure strain (%) Fatigue, 10 s hold 117 219 57.4 0.24 Fatigue, 10 s hold 138 216 55.9 0.29 Fatigue, 100 s hold 128 212 50.7 1.14 Fatigue, 100 s hold 112 219 56.0 1.18 Cyclic–static 128 217 59.4 1.04 0 50 100 150 200 250 0.01 0.1 1 10 100 1000 Time (h) Stress (MPa) As-Processed, Ruggles-Wrenn 2006 Pre-Fatigued UTS T = 1200˚C, Air Fig. 6. Creep stress vs time to rupture for NextelTM720/Alumina ceramic composite at 1200 C in laboratory air. Creep data from Ruggles-Wrenn et al. [27]. 0.0 0.2 0.4 0.6 0.8 1.0 012345 Time (h) Strain (%) T = 1200 o C, Air As-Processed, 125 MPa Ruggles-Wrenn, 2006 As-Processed, 154 MPa Ruggles-Wrenn, 2006 Pre-Fatigued 125 MPa Pre-Fatigued 154 MPa 0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Time (h) Strain (%) T = 1200o C, Steam Creep Stress = 100 MPa As-Processed Ruggles-Wrenn, 2006 Pre-Fatigued Fig. 7. Creep strain vs time for NextelTM720/Alumina ceramic composite at 1200 C: (a) in laboratory air, (b) in steam environment. Creep data from Ruggles-Wrenn et al. [27] are also shown. 1430 J.M. Mehrman et al. / Composites Science and Technology 67 (2007) 1425–1438
J.M. Ehrman et al. Composites Science and Technology 67(2007)1425-1438 10E02 likely due to the porous matrix. Fatigue cycling promotes BAs-Processed, Ruggles-Wrenn 2006 additional matrix cracking as evidenced by a decrease in 1.0E-0 stifness observed in fatigue run-out specimens. In addition, fatigue cycling likely causes the weakening of the fiber matrix interfaces as manifested by a strongly uncorrelated fiber fracture and brushy fracture surfaces produced in fati 1.0E-0 gue in air. It appears that prior fatigue in air serves to improve the damage tolerance and long-term durability 1.0E07 of the N720/A composite as demonstrated by improved creep resistance of this CMC. Note that prior fatigue in T= 1200C. Air steam does not result in similar improvements. It is possible that in steam the beneficial effects of cyclic loading weighed by negative effects of the additional sintering of the matrix. Considerably less brushy fracture surfaces pro- Fig8. Minimum creep rate vs applied stress for N720/A CMC at 1200 c duced in fatigue in steam would support this conjecture. in air. Creep data from Ruggles- Wrenn et al. [27] are also shown. 3. 4. Static-cyclic block loadings 1200C. It retained over 100% of its tensile strength and 76% of its modulus(see Table 2). Effects of prior creep on fatigue durability were investi Creep curves obtained in cyclic-static tests conducted in gated in static-cyclic block loading tests. The maximum ir and in steam are presented in Figs. 7a and b, respec- stress levels were 125 MPa in air and 100 MPa in steam tively. In air, prior fatigue causes a noticeable reduction The creep period was 2 h in air and 0.75 h in steam. In in accumulated creep strains. At 154 MPa, the as-processed air, the pre-crept specimen achieved a fatigue run-out dem pecimen accumulated creep strain of 0.61%o, nearly 3 times onstrating that 2 h of prior creep accompanied with the the 0. 22% creep strain produced by the pre-fatigued speci- creep strain of M0. 28% had no effect on fatigue life up to men. At 125 MPa, the pre-fatigued specimen achieved a run-out, accumulating creep strain of 0.60% in 100 h, 1.00 while the as-processed specimen accumulated 0.55% strain T=1200C. Air in only 0.27 h. In addition, it is seen in Fig. 8 that for a Fatigue Stress 125 MPa given creep stress, creep strain rates for the pre-fatigued specimens are at least an order of magnitude lower than those for the as-processed specimens, reflecting the benefi- cial effect of prior fatigue on the in-air creep performance. 5 0.50 -As-Processed, Ruggles. Wrenn 2006 However, in steam prior fatigue degraded the creep resis- tance. The pre-fatigued specimen failed after only 0.61 h having accumulated creep strain of 0.7%o, while the as-pro- cessed specimen survived 2.5 h producing creep strain of 1.41% 门720/A derives its damage tolerance 0.00 from a porous matrix. Therefore, the stability of the matrix 101520 porosity against densification is critical to the composite's a Time(h) long-term durability. Recent studies [32-34] investigated effects of thermal aging on the physical and mechanical T=1200C. Steam properties of composites consisting of NextelM720 fibers Fatigue Stress= 100 MPa and a porous matrix of mullite and alumina. For a com 0.75 posite with a pure alumina matrix, a porosity reduction of 6% was observed after a 10-min exposure at 1200C strengthening of the matrix and the fiber-matrix interfaces 850 [33, 34]. For a composite with a mullite/alumina matrix. was observed following aging at 1200C [32]. Additional 0.25 As-Processed sintering of the matrix during the aging treatments was Ruggles-Wrenn, 2006 considered to be associated predominantly with Al2O3. among the main manifestations was the increase in the 0.00 spatial correlation in the fiber failure locations within an 101520 individual tow and the increased amount of matrix mate- rial bonded to the fibers. The excellent fatigue resistance Fig 9. Maximum strain vs time for fatigue tests at 1200.C: (a)in air and of the N720/A composite at 1200C in air [27] is most (b)in steam Fatigue data from Ruggles-Wrenn et al. [27] are also shown
1200 C. It retained over 100% of its tensile strength and 76% of its modulus (see Table 2). Creep curves obtained in cyclic–static tests conducted in air and in steam are presented in Figs. 7a and b, respectively. In air, prior fatigue causes a noticeable reduction in accumulated creep strains. At 154 MPa, the as-processed specimen accumulated creep strain of 0.61%, nearly 3 times the 0.22% creep strain produced by the pre-fatigued specimen. At 125 MPa, the pre-fatigued specimen achieved a run-out, accumulating creep strain of 0.60% in 100 h, while the as-processed specimen accumulated 0.55% strain in only 0.27 h. In addition, it is seen in Fig. 8 that for a given creep stress, creep strain rates for the pre-fatigued specimens are at least an order of magnitude lower than those for the as-processed specimens, reflecting the benefi- cial effect of prior fatigue on the in-air creep performance. However, in steam prior fatigue degraded the creep resistance. The pre-fatigued specimen failed after only 0.61 h having accumulated creep strain of 0.7%, while the as-processed specimen survived 2.5 h producing creep strain of 1.41%. The N720/A composite derives its damage tolerance from a porous matrix. Therefore, the stability of the matrix porosity against densification is critical to the composite’s long-term durability. Recent studies [32–34] investigated effects of thermal aging on the physical and mechanical properties of composites consisting of NextelTM720 fibers and a porous matrix of mullite and alumina. For a composite with a pure alumina matrix, a porosity reduction of 6% was observed after a 10-min exposure at 1200 C [33,34]. For a composite with a mullite/alumina matrix, strengthening of the matrix and the fiber–matrix interfaces was observed following aging at 1200 C [32]. Additional sintering of the matrix during the aging treatments was considered to be associated predominantly with Al2O3. Among the main manifestations was the increase in the spatial correlation in the fiber failure locations within an individual tow and the increased amount of matrix material bonded to the fibers. The excellent fatigue resistance of the N720/A composite at 1200 C in air [27] is most likely due to the porous matrix. Fatigue cycling promotes additional matrix cracking as evidenced by a decrease in stiffness observed in fatigue run-out specimens. In addition, fatigue cycling likely causes the weakening of the fiber– matrix interfaces as manifested by a strongly uncorrelated fiber fracture and brushy fracture surfaces produced in fatigue in air. It appears that prior fatigue in air serves to improve the damage tolerance and long-term durability of the N720/A composite as demonstrated by improved creep resistance of this CMC. Note that prior fatigue in steam does not result in similar improvements. It is possible that in steam the beneficial effects of cyclic loading are outweighed by negative effects of the additional sintering of the matrix. Considerably less brushy fracture surfaces produced in fatigue in steam would support this conjecture. 3.4. Static–cyclic block loadings Effects of prior creep on fatigue durability were investigated in static–cyclic block loading tests. The maximum stress levels were 125 MPa in air and 100 MPa in steam. The creep period was 2 h in air and 0.75 h in steam. In air, the pre-crept specimen achieved a fatigue run-out demonstrating that 2 h of prior creep accompanied with the creep strain of 0.28% had no effect on fatigue life up to 1.0E-09 1.0E-08 1.0E-07 1.0E-06 1.0E-05 1.0E-04 1.0E-03 1.0E-02 10 100 1000 Creep Stress (MPa) Creep Strain Rate (s-1) As-Processed, Ruggles-Wrenn 2006 Pre-Fatigued T = 1200˚C, Air Fig. 8. Minimum creep rate vs applied stress for N720/A CMC at 1200 C in air. Creep data from Ruggles-Wrenn et al. [27] are also shown. 0.00 0.25 0.50 0.75 1.00 0 5 10 15 20 25 30 Time (h) Strain (%) As-Processed, Ruggles-Wrenn 2006 2 h at 125 MPa T = 1200 ˚C, Air Fatigue Stress = 125 MPa 0.00 0.25 0.50 0.75 1.00 0 5 10 15 20 25 30 Time (h) Strain (%) T = 1200˚C, Steam Fatigue Stress = 100 MPa As-Processed Ruggles-Wrenn, 2006 0.75 h at 100 MPa Fig. 9. Maximum strain vs time for fatigue tests at 1200 C: (a) in air and (b) in steam. Fatigue data from Ruggles-Wrenn et al. [27] are also shown. J.M. Mehrman et al. / Composites Science and Technology 67 (2007) 1425–1438 1431
J M. Ehrman et al Composites Science and Technology 67(2007)1425-1438 105 cycles. Furthermore, as seen in Fig. 9a, where maxi- maximum strain vs time curves in Fig. 9b show that the mum strain is plotted vs time for fatigue tests in air, strains pre-crept specimen accumulates strain rapidly, reaching accumulated by the pre-crept specimen were small (<0.1%) failure strain of 0. 75% in 14.6 h, while the as-processed and close to those produced by the as-processed material. specimen accumulates less than 0.5% strain in 28 h However in steam, 0.75 h of prior creep accompanied with a significant creep strain of 0.9% considerably degraded 3. 5. Composite microstructure fatigue performance. For the stress of 100 MPa, prior creep reduced the fatigue life by a factor of 2. The pre-crept spec Optical micrographs of fracture surfaces obtained in imen failed after 52, 620 cycles(146 h), while the as-pro- 125 MPa fatigue tests with 10-s hold conducted in air and cessed specimen achieved a run-out of 10 cycles. The in steam are shown in Figs. 10a and b, respectively. The (a) (b) 2mm kmm Fig. 10. Fracture surfaces obtained in 125 MPa fatigue tests with 10-s hold conducted at 1200.C: (a)in air(specimen achieved a run-out)and (b)in steam b 1. 0 mm 100pm 1123030AM ewm Mso ow l i sot d 10 um Fig. Il. Fracture surfaces of N720/A specimens tested at 1200C:(a)overall view showing fiber pullout and regions of coordinated fracture, (b) individual fiber pullout, (c) nearly planar fracture in the 0 tow, (d) matrix particles bonded to the fibe
105 cycles. Furthermore, as seen in Fig. 9a, where maximum strain is plotted vs time for fatigue tests in air, strains accumulated by the pre-crept specimen were small (<0.1%) and close to those produced by the as-processed material. However in steam, 0.75 h of prior creep accompanied with a significant creep strain of 0.9% considerably degraded fatigue performance. For the stress of 100 MPa, prior creep reduced the fatigue life by a factor of 2. The pre-crept specimen failed after 52,620 cycles (14.6 h), while the as-processed specimen achieved a run-out of 105 cycles. The maximum strain vs time curves in Fig. 9b show that the pre-crept specimen accumulates strain rapidly, reaching failure strain of 0.75% in 14.6 h, while the as-processed specimen accumulates less than 0.5% strain in 28 h. 3.5. Composite microstructure Optical micrographs of fracture surfaces obtained in 125 MPa fatigue tests with 10-s hold conducted in air and in steam are shown in Figs. 10a and b, respectively. The Fig. 10. Fracture surfaces obtained in 125 MPa fatigue tests with 10-s hold conducted at 1200C: (a) in air (specimen achieved a run-out) and (b) in steam (tf = 0.23 h). Fig. 11. Fracture surfaces of N720/A specimens tested at 1200 C: (a) overall view showing fiber pullout and regions of coordinated fracture, (b) individual fiber pullout, (c) nearly planar fracture in the 0 tow, (d) matrix particles bonded to the fiber. 1432 J.M. Mehrman et al. / Composites Science and Technology 67 (2007) 1425–1438
J.M. Ehrman et aL/Ce tes Science and Technology 67(2007)1425-1438 1433 fracture planes of both specimens are not well defined. The which shows matrix material still bonded to the pulled 0o fiber tows break over a wide range of axial locations in out fiber general spanning the entire width of the specimen. The While regions of both brushy and nearly planar failure fibers in the 0 tows in each cloth layer exhibit random fail- are present in all fracture surfaces, the balance of these ure producing brushy fracture surfaces. Note that the spec- two fracture topographies within a given fracture surface imen tested in air, which achieved a fatigue run-out and is influenced by the test type(see Fig 12). The fracture sur failed in the subsequent tensile test, has a considerably face produced in fatigue(Fig. 12a)is dominated by areas of longer damage zone than the specimen tested in steam, uncorrelated fiber fracture where individual fibers are produced damage zones ranging from I to 13 mm in in fatigue with 10 s hold(Fig. 12b) still exhibits multiple length. It is noteworthy that specimens which exhibited areas of uncoordinated brushy failure, areas of planar frac- longer lifetimes invariably produced longer damage zones. ture are also visible. Coordinated fracture becomes more All fracture surfaces obtained in this effort contain bru- prevalent as the hold time increases to 100s(Fig. 12c) shy regions of fibrous fracture as well as regions of flatter, Finally, the fracture surface produced in creep(Fig. 12d) more coordinated fracture topography. Typical features of is dominated by areas of nearly planar failure. It is seen the composite microstructure are shown in Fig. ll. An that extensive fiber pullout is produced in cyclic loading overall view of a fracture surface is presented in Fig. lla. and predominantly planar fracture, in creep. Once a hold Fig. Ilb shows pullout of individual fibers. Note that the time is introduced into a fatigue cycle, regions of coordi- locations of the fiber failure within an individual tow, nated failure are seen. The planar fracture topography and consequently the lengths of fiber pullout exhibit a becomes more prevalent with increasing hold time broad distribution Region of coordinated fracture of the Fracture surfaces obtained in the 125 MPa creep-fatigue 0o tows is seen in Fig. llc. Coordinated fiber fracture typ- interaction tests in steam are shown in Fig. 13. The fatigue ally is indicative of a single crack front passing through fracture surface exhibits mostly uncorrelated fiber failure. the tow. Evidence of the strong fiber-matrix bond, charac- while the creep fracture surface is almost entirely planar teristic of the present composite can be seen in Fig. Ild, However, the fracture surfaces produced in fatigue with 如mM 1.0mm Fig. 12. Fracture surfaces of the N720/A CMC tested at 1200C in air with omax= 125 MPa: (a)fatigue, (b )fatigue with 10-s hold, (c) fatigue with 100-s hold, and (d) creep
fracture planes of both specimens are not well defined. The 0 fiber tows break over a wide range of axial locations, in general spanning the entire width of the specimen. The fibers in the 0 tows in each cloth layer exhibit random failure producing brushy fracture surfaces. Note that the specimen tested in air, which achieved a fatigue run-out and failed in the subsequent tensile test, has a considerably longer damage zone than the specimen tested in steam, which failed after 0.23 h. Specimens tested in this effort produced damage zones ranging from 1 to 13 mm in length. It is noteworthy that specimens which exhibited longer lifetimes invariably produced longer damage zones. All fracture surfaces obtained in this effort contain brushy regions of fibrous fracture as well as regions of flatter, more coordinated fracture topography. Typical features of the composite microstructure are shown in Fig. 11. An overall view of a fracture surface is presented in Fig. 11a. Fig. 11b shows pullout of individual fibers. Note that the locations of the fiber failure within an individual tow, and consequently the lengths of fiber pullout exhibit a broad distribution. Region of coordinated fracture of the 0 tows is seen in Fig. 11c. Coordinated fiber fracture typically is indicative of a single crack front passing through the tow. Evidence of the strong fiber–matrix bond, characteristic of the present composite can be seen in Fig. 11d, which shows matrix material still bonded to the pulled out fiber. While regions of both brushy and nearly planar failure are present in all fracture surfaces, the balance of these two fracture topographies within a given fracture surface is influenced by the test type (see Fig. 12). The fracture surface produced in fatigue (Fig. 12a) is dominated by areas of uncorrelated fiber fracture where individual fibers are clearly discernable. While the fracture surface produced in fatigue with 10 s hold (Fig. 12b) still exhibits multiple areas of uncoordinated brushy failure, areas of planar fracture are also visible. Coordinated fracture becomes more prevalent as the hold time increases to 100 s (Fig. 12c). Finally, the fracture surface produced in creep (Fig. 12d) is dominated by areas of nearly planar failure. It is seen that extensive fiber pullout is produced in cyclic loading and predominantly planar fracture, in creep. Once a hold time is introduced into a fatigue cycle, regions of coordinated failure are seen. The planar fracture topography becomes more prevalent with increasing hold time. Fracture surfaces obtained in the 125 MPa creep–fatigue interaction tests in steam are shown in Fig. 13. The fatigue fracture surface exhibits mostly uncorrelated fiber failure, while the creep fracture surface is almost entirely planar. However, the fracture surfaces produced in fatigue with Fig. 12. Fracture surfaces of the N720/A CMC tested at 1200 C in air with rmax = 125 MPa: (a) fatigue, (b) fatigue with 10-s hold, (c) fatigue with 100-s hold, and (d) creep. J.M. Mehrman et al. / Composites Science and Technology 67 (2007) 1425–1438 1433
J M. Ehrman et al Composites Science and Technology 67(2007)1425-1438 b 23p% 10 V mm too, 2 OK d 1.0mm Fig 13. Fracture surfaces of the N720/A CMC tested at 1200C in steam with Omax=125 MPa: (a)fatigue, (b) fatigue with 10-s hold, (c) fatigue with 100-s hold, and(d) creep hold time are also dominated by regions of planar fracture. high-temperature creep performance of the Nextel720 The hold duration appears to have little effect on the frac- fibers is due to the high content of mullite, which has a ture topography. Once a 10-s hold is introduced, planar much better creep resistance than alumina [35, 36]. Creep fracture prevails. Increasing the hold to 100 s has little resistance of NextelmM720 fibers is also due in part to the lar dditional effect. Note that the specimen tested in fatigue ger grain size of this fiber relative to other oxide fibers such ig 13a), which produced the largest amount of fiber pull- as Nextel 610 [35, 36]. A recent study [37]concluded that out, also achieved a run-out. On the contrary, specimens at 1100C in water-vapor environment, Sioz could be lea- with nearly planar fracture surfaces shown in Figs. 13b- ched from NextelTM720 fiber. Wannaparhun et al. [37]sug- d,exhibited short lives of 0. 22-0. h In air as well as in gested that exposure of the N720/A CMC to water vapor steam,planar fracture surface accompanies short time to at 1100C could cause an increase in Al2O3 content of failure, while brushy fracture surface is indicative of longer the fiber due to the loss of Sio2 from its mullite phase fe. For both air and steam environments, the fracture sur- Campbell et al. [38]observed a M15% reduction in strength faces obtained in fatigue are readily distinguished from of the N720/A composite after a 1000-h exposure to water those produced in creep. In addition, the fracture surface vapor environment at 1200C appearance may be correlated with time to failure a qualitative EDS analysis was performed to assess whether the loss of the mullite phase in the fiber may be 3.6. Energy dispersive X-ray spectroscopy(EDS) the mechanism behind the reduced creep resistance in steam. Specimens subjected to fatigue with 10-s hold at Results of the present study as well as the prior work 125 MPa in air and in steam as well as the as-processed [27]reveal the degrading effect of steam on fatigue as well material were examined. The electron beam spot mode as on creep performance of the N720/A composite at was employed in order to precisely control the beam place- 1200C. Mechanical behavior of a composite with 0/90 ment. Due to the uncertainties associated with the size and fiber orientation under tensile loading is fiber-dominated. shape of the reaction volume, the EDS results are inter Therefore, fiber degradation represents a likely source of preted as qualitative. All data were normalized to a CPS composite degradation. It is recognized that the superior of 1.0 for aluminum
hold time are also dominated by regions of planar fracture. The hold duration appears to have little effect on the fracture topography. Once a 10-s hold is introduced, planar fracture prevails. Increasing the hold to 100 s has little additional effect. Note that the specimen tested in fatigue (Fig. 13a), which produced the largest amount of fiber pullout, also achieved a run-out. On the contrary, specimens with nearly planar fracture surfaces shown in Figs. 13b– d, exhibited short lives of 0.22–0.31 h. In air as well as in steam, planar fracture surface accompanies short time to failure, while brushy fracture surface is indicative of longer life. For both air and steam environments, the fracture surfaces obtained in fatigue are readily distinguished from those produced in creep. In addition, the fracture surface appearance may be correlated with time to failure. 3.6. Energy dispersive X-ray spectroscopy (EDS) Results of the present study as well as the prior work [27] reveal the degrading effect of steam on fatigue as well as on creep performance of the N720/A composite at 1200 C. Mechanical behavior of a composite with 0/90 fiber orientation under tensile loading is fiber-dominated. Therefore, fiber degradation represents a likely source of composite degradation. It is recognized that the superior high-temperature creep performance of the NextelTM720 fibers is due to the high content of mullite, which has a much better creep resistance than alumina [35,36]. Creep resistance of NextelTM720 fibers is also due in part to the larger grain size of this fiber relative to other oxide fibers such as NextelTM610 [35,36]. A recent study [37] concluded that at 1100 C in water-vapor environment, SiO2 could be leached from NextelTM720 fiber. Wannaparhun et al. [37] suggested that exposure of the N720/A CMC to water vapor at 1100 C could cause an increase in Al2O3 content of the fiber due to the loss of SiO2 from its mullite phase. Campbell et al. [38] observed a 15% reduction in strength of the N720/A composite after a 1000-h exposure to watervapor environment at 1200 C. A qualitative EDS analysis was performed to assess whether the loss of the mullite phase in the fiber may be the mechanism behind the reduced creep resistance in steam. Specimens subjected to fatigue with 10-s hold at 125 MPa in air and in steam as well as the as-processed material were examined. The electron beam spot mode was employed in order to precisely control the beam placement. Due to the uncertainties associated with the size and shape of the reaction volume, the EDS results are interpreted as qualitative. All data were normalized to a CPS of 1.0 for aluminum. Fig. 13. Fracture surfaces of the N720/A CMC tested at 1200 C in steam with rmax = 125 MPa: (a) fatigue, (b) fatigue with 10-s hold, (c) fatigue with 100-s hold, and (d) creep. 1434 J.M. Mehrman et al. / Composites Science and Technology 67 (2007) 1425–1438