Composites Science and Technology 69(2009)890-897 Contents lists available at Science Direct Composites Science and Technology ELSEVIER journalhomepagewww.elsevier.com/locate/compscitech Elevated-temperature stress rupture in interlaminar shear of a Hi-Nic SiC/Sic ceramic matrIx composite Sung R. Choi, Robert W. Kowalik, Donald ]. Alexander a, Narottam P Bansal b - Naval Air Systems Command. 48066 Shaw Road, Patuxent River, MD 20670, United States National Aeronautics and Space Administration, Glenn Research Center, Cleveland, OH 44135, United States ARTICLE INF O A BSTRACT Article history: Assessments of stress rupture in a gas-turbine grade, melt-infiltrated Hi-Nicalon"M SiC continuous fiber Received 1 August 2008 reinforced SiC ceramic matrix composite(CMC)were made in interlaminar shear at 1316C in air. The ccepted 10 December 2008 composite exhibited appreciable life limiting behavior with a life susceptibility parameter of ns-22 Available online 24 December 2008 24, estimated based on a proposed phenomenological life prediction model together with experiment data. The phenomenological life model was in good agreement in prediction between the stress rupture and the constant stress-rate data, validating its appropriateness in describing the life limiting non of the CMC coupons subjected laminar shear. The results of this work also indicate B Mechanical properties governing mechanism(s) associated with failure in interlaminar shear would have remaine unchanged, regardless of the type of loading configurations, either in stress rupture or in constant stress C. Life prediction Published by Elsevier Ltd. D Stress rupture 1 Introduction to improve interlaminar properties of CMCs have been made using 3D preform architectures such as orthogonal and angle-interlock The successful development and design of continuous fiber- weaving, typically with some expense of their in-plane properties. reinforced ceramic matrix composites(CMCs)are dependent on However, the issues on affordability, reliability and manufactura- better understanding of their life limiting properties such as fati- bility of those 3D reinforced CMCs are still in question and need gue, slow crack growth, stress rupture, creep, and environmental to be answered. degradation. Life limiting phenomenon of CMCs can occur individ- Efforts regarding the assessments of life limiting properties of ually or in combination of some of these limiting properties CMCs have been focused mainly on the in-plane direction. However, depending on temperature and environment. Accurate and reliable because of their inherent architectural features, CMCs would exhibit evaluation of life limiting behavior under specific loading/temper- life limiting behavior in interlaminar shear or tension at elevated ture/environmental conditions is important to ensure accurate temperatures. Presence of such a life limiting aspect in addition to life prediction of structural CMC components heirinferior interlaminar properties would make CMC systems sig- Although fiber-reinforced CMCs have shown much nificant drawbacks/weakness in structural integrity, particularly resistance to fracture and increased damage tolerance in when used as aero-propulsion components. In a previous study direction, inherent material/processing defects, voi 6. the life limiting properties of a cross-plied glass ceramic com- cracks in the matrix-rich or fiber-matrix interface regions can still posite(Hi-Nic SiC fiber-reinforced barium strontium aluminosili cause delamination under interlaminar tensile or shear stress, cate matrix composite, SiC /BSAS) were evaluated in shear at resulting in loss of stiffness or in some cases structural failure. 1100C in air by using double notch shear test specimens subjected Interlaminar tensile and shear strength behaviors of CMCs have to constant shear stress-rate loading. The composite exhibited shear been characterized in view of their unique interfacial architectures strength degradation with decreasing stress rates, analogous to the and importance in structural ap [1-4. It has been re- slow crack growth(SCG) process occurring in tension for many ported that many 2D woven SiC/SiC CMCs exhibited poor interlam ceramics at elevated temperatures. The life limiting susceptibility lar properties with interlaminar shear strength of 30-50 MPa and was significant with a life susceptibility parameter(n)of about 10. interlaminar tensile strength of 10-20 MPa 5]. Various attempts The life of the CMc in shear was limited by slow crack growth SCG)or damage accumulation/growth. The life limiting behavior Corresponding author. Tel :+1 301 342 8074; fax: +1 301 342 8044 was modeled using a power-law type of pseudo-SCG process of a crack located at matrix-rich interfaces. This model has been applied ront matter Published by Elsevier Ltd
Elevated-temperature stress rupture in interlaminar shear of a Hi-Nic SiC/SiC ceramic matrix composite Sung R. Choi a,*, Robert W. Kowalik a , Donald J. Alexander a , Narottam P. Bansal b aNaval Air Systems Command, 48066 Shaw Road, Patuxent River, MD 20670, United States bNational Aeronautics and Space Administration, Glenn Research Center, Cleveland, OH 44135, United States article info Article history: Received 1 August 2008 Received in revised form 21 October 2008 Accepted 10 December 2008 Available online 24 December 2008 Keywords: A. Ceramic matrix composite B. Mechanical properties A. MI SiC/SiC D. Interlaminar shear C. Life prediction D. Stress rupture abstract Assessments of stress rupture in a gas-turbine grade, melt-infiltrated Hi-NicalonTM SiC continuous fiberreinforced SiC ceramic matrix composite (CMC) were made in interlaminar shear at 1316 C in air. The composite exhibited appreciable life limiting behavior with a life susceptibility parameter of ns = 22– 24, estimated based on a proposed phenomenological life prediction model together with experimental data. The phenomenological life model was in good agreement in prediction between the stress rupture and the constant stress-rate data, validating its appropriateness in describing the life limiting phenomenon of the CMC coupons subjected to interlaminar shear. The results of this work also indicated that the governing mechanism(s) associated with failure in interlaminar shear would have remained almost unchanged, regardless of the type of loading configurations, either in stress rupture or in constant stress rate. Published by Elsevier Ltd. 1. Introduction The successful development and design of continuous fiberreinforced ceramic matrix composites (CMCs) are dependent on better understanding of their life limiting properties such as fatigue, slow crack growth, stress rupture, creep, and environmental degradation. Life limiting phenomenon of CMCs can occur individually or in combination of some of these limiting properties depending on temperature and environment. Accurate and reliable evaluation of life limiting behavior under specific loading/temperature/environmental conditions is important to ensure accurate life prediction of structural CMC components. Although fiber-reinforced CMCs have shown much improved resistance to fracture and increased damage tolerance in in-plane direction, inherent material/processing defects, voids, and/or cracks in the matrix-rich or fiber–matrix interface regions can still cause delamination under interlaminar tensile or shear stress, resulting in loss of stiffness or in some cases structural failure. Interlaminar tensile and shear strength behaviors of CMCs have been characterized in view of their unique interfacial architectures and importance in structural applications [1–4]. It has been reported that many 2D woven SiC/SiC CMCs exhibited poor interlaminar properties with interlaminar shear strength of 30–50 MPa and interlaminar tensile strength of 10–20 MPa [5]. Various attempts to improve interlaminar properties of CMCs have been made using 3D preform architectures such as orthogonal and angle-interlock weaving, typically with some expense of their in-plane properties. However, the issues on affordability, reliability and manufacturability of those 3D reinforced CMCs are still in question and need to be answered. Efforts regarding the assessments of life limiting properties of CMCs have been focused mainly on the in-plane direction. However, because of their inherent architectural features, CMCs would exhibit life limiting behavior in interlaminar shear or tension at elevated temperatures. Presence of such a life limiting aspect in addition to their inferior interlaminar properties would make CMC systems significant drawbacks/weakness in structural integrity, particularly when used as aero-propulsion components. In a previous study [6], the life limiting properties of a cross-plied glass ceramic composite (Hi-Nic SiC fiber-reinforced barium strontium aluminosilicate matrix composite, SiC/BSAS) were evaluated in shear at 1100 C in air by using double notch shear test specimens subjected to constant shear stress-rate loading. The composite exhibited shear strength degradation with decreasing stress rates, analogous to the slow crack growth (SCG) process occurring in tension for many ceramics at elevated temperatures. The life limiting susceptibility was significant with a life susceptibility parameter (n) of about 10. The life of the CMC in shear was limited by slow crack growth (SCG) or damage accumulation/growth. The life limiting behavior was modeled using a power-law type of pseudo-SCG process of a crack located at matrix-rich interfaces. This model has been applied 0266-3538/$ - see front matter Published by Elsevier Ltd. doi:10.1016/j.compscitech.2008.12.006 * Corresponding author. Tel.: +1 301 342 8074; fax: +1 301 342 8044. E-mail address: sung.choi1@navy.mil (S.R. Choi). Composites Science and Technology 69 (2009) 890–897 Contents lists available at ScienceDirect Composites Science and Technology journal homepage: www.elsevier.com/locate/compscitech
S.R. Chof et al / Composites Science and Technology 69(2009)890-897 891 to other CMCs such as SiC/SiCs, SiC/MAS(magnesium aluminosili- to poor CVl, slurry, and or MI infiltration processes. Porosity was cate), and C/Sic [7. Recently, creep behavior in interlaminar shear estimated to be approximately 10%. has been assessed for an oxide/oxide( Nextel720/alumina)CMC in and steam environments by Ruggles-Wrenn 2. 2. Stress rupture testing This paper, as an extension of the previous work [6,7]. describes ife limiting behavior of a commercial, gas-turbine grade, melt Stress rupture testing for the Hi-Nic SiC/SiC composite was con- infiltrated(Mi) Hi-Nic SiC fiber-reinforced Sic matrix composite ducted in interlaminar shear at 1316C in air. The double notch (designated Hi-Nic SiC/SiC) in interlaminar shear. Double notch shear(DNS) test specimens were machined from one of the com- shear(DNS)test specimens using a significant sample size were posite panels. Test specimens were 12.7 mm wide (w) and tested at 1316C in air under stress rupture loading. Life limiting 30 mm long(L). The thickness of test specimens corresponded to behavior of the Hi-Nic SiC/SiC composite was analyzed using a the nominal thickness(=2 mm) of the composite panels. Two power-law type of phenomenological life model proposed previ- notches, 0.3 mm wide(h)and 6 mm (Ln)away from each other ously [6,7. Additional experimental data, determined in constant were made into each test specimen such that the two notches were shear stress-rate testing at 1316C in air, were also used to further extended to the middle of the specimen so that shear failure oc- validate the proposed model. Some of the work has been reported curred on the plane between the notch tips. Schematics of a DNS previously [9] and more detailed work can be found in a recent re- test specimen and the test setup are shown in Fig. 2. Test fixtures port [10]. yere all made of a-SiC. In-situ axial displacement of test speci mens was monitored for some selected samples using an LVDt 2. Experimental procedures placed between the upper and lower fixtures, incorporated with a data acquisition system. Since test specimens were relatively thin 21. Material (2 mm)and somewhat irregular in their surfaces due to the nature of 2D fabrication, specially designed Sic anti-buckling guides were A 2D woven Hi-NicalonTM sic fiber-reinforced Sic ceramic ma- used. Each test specimen was kept prior to testing for about 20 min trix composite(Hi-Nic SiC/SiC) fabricated by GE Power System at 1316C for thermal equilibration. A total of 22 test specimens Composites(Newark, DL; vintage 02), was used in this study De- were tested over a total of five different levels of applied shear tailed descriptions of the composite and its processing can be stresses ranging from 8 to 17. 8 MPa Test specimen configurations found elsewhere [11]. Briefly, Hi- Nic SiC fibers, produced in tow, and test procedure were followed in accordance with ASTM C1425 were woven into 2D 5 harness satin fabric at Techni-Weave(al [14. All testing was performed using an electromechanical test bany, NY). The fabric preforms of the compos osite were cut into frame(Model 8562, Instron, Canton, MA). Time to failure of each 200 mm x 150 mm, eight ply-sta hemically vapor infil- specimen tested was determined from the data acquisition system trated(CVI)with a thin BN-based interface coating followed by Interlaminar shear stress, i.e the average nominal shear stress. SiC matrix over-coating Remaining matrix porosity was filled with was calculated using the following relation iC particulates and then with molten silicon at 1400C, a process termed slurry casting and melt infiltration (Mi). The MI SiC/Sic - WL (1) composite was composed of about 39 vol% fibers, about 8 vol% BN coating, about 25 vol% SiC coating, and about 28 vol% SiC partic- where t is the applied shear stress, P is the applied load(in com- ulates, silicon and pores. The nominal dimensions of the composite pression), and w and Ln are the specimen width and the distance panels fabricated were about 200 mm by 150 mm with a thickness between the two notches, respectively (see Fig. 2). of about 2.0 mm. The composite exhibited 300 MPa in-plane ten- sile strength, 36+3 MPa interlaminar shear strength, 13+1 MPa 23. Constant shear stress-rate testing transthickness tensile strength, 183 +7 GPa in-plane elastic modu lus(determined by the impulse excitation of vibration technique, Additional interlaminar shear testing was also conducted using ASTM C1259 [12)) and 2.36+0.02 g/cm bulk density, all at ambi- constant shear stress-rate testing Each test specimen, cut from the ent temperature [5). Mode I and ll interlaminar crack growth resis- same composite panel that was used in stress rupture, was sub tances were determined at ambient temperature as Gr Gu 200- jected to a given applied shear stress rate until it failed. A total 600J/m2[13]. A typical micrograph of the cross section of the com- of three different shear stress rates ranging from 5 to 0.005 MPa/ posite is shown in Fig. 1. Porosity was evident with some signifi- s were employed with a total of five specimens tested at each ap- cant voids in tows and matrices which were probably attributed lied shear stress rate. One test specimen was tested at 0.0005 MPa/s Test fixtures, test temperature, test specimen config uration, and test frame were the same as those used in stress rup- ture testing. This type of testing, in which strength is determined as a function of stress rate. is often called 'constant stress-rate dynamic fatigue testing when applied to glasses and monolithic ceramics to determine their slow crack growth (SCG) behavior in tension or flexure [15, 16. The objective of this supplementary testing was to determine life limiting behavior in constant stress- rate loading and to compare it with that in stress rupture, with which the phenomenological life prediction model can be val has been utilized to determine time-dependent shear failure behavior of some CMCs such as SiC/SiCs, SIC/MAS, C/Sic [7, and 500um SiC/BSAS [6]. Applied shear stress rate (t) was calculated using Fig. 1. Microstructure of 2D, 5HS, melt-infiltrated (MI) Hi-Nic Sic/ Sic ceramic t matrx co used in this work
to other CMCs such as SiC/SiCs, SiC/MAS (magnesium aluminosilicate), and C/SiC [7]. Recently, creep behavior in interlaminar shear has been assessed for an oxide/oxide (Nextel720/alumina) CMC in air and steam environments by Ruggles-Wrenn [8]. This paper, as an extension of the previous work [6,7], describes life limiting behavior of a commercial, gas-turbine grade, meltinfiltrated (MI) Hi-Nic SiC fiber-reinforced SiC matrix composite (designated Hi-Nic SiC/SiC) in interlaminar shear. Double notch shear (DNS) test specimens using a significant sample size were tested at 1316 C in air under stress rupture loading. Life limiting behavior of the Hi-Nic SiC/SiC composite was analyzed using a power-law type of phenomenological life model proposed previously [6,7]. Additional experimental data, determined in constant shear stress-rate testing at 1316 C in air, were also used to further validate the proposed model. Some of the work has been reported previously [9] and more detailed work can be found in a recent report [10]. 2. Experimental procedures 2.1. Material A 2D woven Hi-NicalonTM SiC fiber-reinforced SiC ceramic matrix composite (Hi-Nic SiC/SiC), fabricated by GE Power System Composites (Newark, DL; vintage ’02), was used in this study. Detailed descriptions of the composite and its processing can be found elsewhere [11]. Briefly, Hi-Nic SiC fibers, produced in tow, were woven into 2D 5 harness satin fabric at Techni-Weave (Albany, NY). The fabric preforms of the composite were cut into 200 mm 150 mm, eight ply-stacked, and chemically vapor infiltrated (CVI) with a thin BN-based interface coating followed by SiC matrix over-coating. Remaining matrix porosity was filled with SiC particulates and then with molten silicon at 1400 C, a process termed slurry casting and melt infiltration (MI). The MI SiC/SiC composite was composed of about 39 vol% fibers, about 8 vol% BN coating, about 25 vol% SiC coating, and about 28 vol% SiC particulates, silicon and pores. The nominal dimensions of the composite panels fabricated were about 200 mm by 150 mm with a thickness of about 2.0 mm. The composite exhibited 300 MPa in-plane tensile strength, 36 ± 3 MPa interlaminar shear strength, 13 ± 1 MPa transthickness tensile strength, 183 ± 7 GPa in-plane elastic modulus (determined by the impulse excitation of vibration technique, ASTM C 1259 [12]), and 2.36 ± 0.02 g/cm3 bulk density, all at ambient temperature [5]. Mode I and II interlaminar crack growth resistances were determined at ambient temperature as GI, GII 200– 600 J/m2 [13]. A typical micrograph of the cross section of the composite is shown in Fig. 1. Porosity was evident with some signifi- cant voids in tows and matrices which were probably attributed to poor CVI, slurry, and/or MI infiltration processes. Porosity was estimated to be approximately 10%. 2.2. Stress rupture testing Stress rupture testing for the Hi-Nic SiC/SiC composite was conducted in interlaminar shear at 1316 C in air. The double notch shear (DNS) test specimens were machined from one of the composite panels. Test specimens were 12.7 mm wide (W) and 30 mm long (L). The thickness of test specimens corresponded to the nominal thickness (=2 mm) of the composite panels. Two notches, 0.3 mm wide (h) and 6 mm (Ln) away from each other, were made into each test specimen such that the two notches were extended to the middle of the specimen so that shear failure occurred on the plane between the notch tips. Schematics of a DNS test specimen and the test setup are shown in Fig. 2. Test fixtures were all made of a-SiC. In-situ axial displacement of test specimens was monitored for some selected samples using an LVDT placed between the upper and lower fixtures, incorporated with a data acquisition system. Since test specimens were relatively thin (2 mm) and somewhat irregular in their surfaces due to the nature of 2D fabrication, specially designed SiC anti-buckling guides were used. Each test specimen was kept prior to testing for about 20 min at 1316 C for thermal equilibration. A total of 22 test specimens were tested over a total of five different levels of applied shear stresses ranging from 8 to 17.8 MPa. Test specimen configurations and test procedure were followed in accordance with ASTM C 1425 [14]. All testing was performed using an electromechanical test frame (Model 8562, Instron, Canton, MA). Time to failure of each specimen tested was determined from the data acquisition system. Interlaminar shear stress, i.e., the average nominal shear stress, was calculated using the following relation: s ¼ P WLn ð1Þ where s is the applied shear stress, P is the applied load (in compression), and W and Ln are the specimen width and the distance between the two notches, respectively (see Fig. 2). 2.3. Constant shear stress-rate testing Additional interlaminar shear testing was also conducted using constant shear stress-rate testing. Each test specimen, cut from the same composite panel that was used in stress rupture, was subjected to a given applied shear stress rate until it failed. A total of three different shear stress rates ranging from 5 to 0.005 MPa/ s were employed with a total of five specimens tested at each applied shear stress rate. One test specimen was tested at 0.0005 MPa/s. Test fixtures, test temperature, test specimen configuration, and test frame were the same as those used in stress rupture testing. This type of testing, in which strength is determined as a function of stress rate, is often called ‘constant stress-rate’ or ‘dynamic fatigue’ testing when applied to glasses and monolithic ceramics to determine their slow crack growth (SCG) behavior in tension or flexure [15,16]. The objective of this supplementary testing was to determine life limiting behavior in constant stressrate loading and to compare it with that in stress rupture, with which the phenomenological life prediction model can be validated. Constant shear stress-rate testing at elevated temperatures has been utilized to determine time-dependent shear failure behavior of some CMCs such as SiC/SiCs, SiC/MAS, C/SiC [7], and SiC/BSAS [6]. Applied shear stress rate (s_) was calculated using the relation s_ ¼ _ P WLn Fig. 1. Microstructure of 2D, 5HS, melt-infiltrated (MI) Hi-Nic SiC/SiC ceramic ð2Þ matrix composite used in this work. S.R. Choi et al. / Composites Science and Technology 69 (2009) 890–897 891
SR Choi et al/ Composites Science and Technology 69(2009)890-897 LVDT. Test specimen Thermocouple Fig. 2.(a) Geometry of double notch shear(DNS) test specimen and (b) schematic showing a test setup used in this work. where P is the applied load rate (in compression ) which can be ap- plied directly to test specimens via a test frame under load control. DNS at 1316c 3. Results 20 3.1. Stress rupture All the specimens tested in stress rupture, except for the two L run out specimens, failed in typical shear mode along their respec tive interlaminar shear planes. An example of such a mode of shear failure is shown in Fig 3. The results of stress rupture testing are 3 presented in Fig. 4, where by convention applied shear stress vas plotted as a function of time to failure. The data show an evi- 4 1min 0h100h1000h dence of life limiting behavior, where time to failure decreased with increasing applied shear stress. The solid line represents the best fit based on the log(time to failure)vs log(applied interlaminar 102 03104105106107 shear stress)relation, which will be discussed later. A relatively failure, t, [s] large scatter in time to failure is noted, similar to the feature ob- served in many CMCs and advanced monolithic ceramics subjected Fig 4. Results of stress rupture testing for Hi-Nic Sic/ SiC composite in interlaminar to stress rupture in tension or flexure at elevated temperatures. shear at 1316C in air. The solid line represents the best fit. The arrowed data points Statistical aspects of the data scatter in stress rupture will be de- scribed in Section 4 in conjunction with those in constant stress rate ure within one or two adjacent layers. Well-preserved imprints of Fracture surfaces of tested specimens in stress rupture(or in fibers on the mating matrices were usually seen from the fracture constant stress-rate tests)showed matrix-rich interface shear fail- surfaces. In addition, some breakage of horizontal (warp)and ver sher stine ane 3. Examples of (a)an as-machined double notch shear(DNS)test specimen and (b)a specimen tested in interlaminar shear stress rupture at 17. 8 MP at 1316"C in air a resulting failure time of 400 s. The material was 2D, 5HS, melt-infiltrated (Mi) Hi-Nic Sic/SiC composit
where _ P is the applied load rate (in compression), which can be applied directly to test specimens via a test frame under load control. 3. Results 3.1. Stress rupture All the specimens tested in stress rupture, except for the two run out specimens, failed in typical shear mode along their respective interlaminar shear planes. An example of such a mode of shear failure is shown in Fig. 3. The results of stress rupture testing are presented in Fig. 4, where by convention applied shear stress was plotted as a function of time to failure. The data show an evidence of life limiting behavior, where time to failure decreased with increasing applied shear stress. The solid line represents the best fit based on the log (time to failure) vs. log (applied interlaminar shear stress) relation, which will be discussed later. A relatively large scatter in time to failure is noted, similar to the feature observed in many CMCs and advanced monolithic ceramics subjected to stress rupture in tension or flexure at elevated temperatures. Statistical aspects of the data scatter in stress rupture will be described in Section 4 in conjunction with those in constant stress rate. Fracture surfaces of tested specimens in stress rupture (or in constant stress-rate tests) showed matrix-rich interface shear failure within one or two adjacent layers. Well-preserved imprints of fibers on the mating matrices were usually seen from the fracture surfaces. In addition, some breakage of horizontal (warp) and verW L Ln t h Upper fixture Test specimen Lower fixture Thermocouple Guides P LVDT Anti-buckling guide a b Fig. 2. (a) Geometry of double notch shear (DNS) test specimen and (b) schematic showing a test setup used in this work. Fig. 3. Examples of (a) an as-machined double notch shear (DNS) test specimen and (b) a specimen tested in interlaminar shear stress rupture at 17.8 MP at 1316 C in air with a resulting failure time of 400 s. The material was 2D, 5HS, melt-infiltrated (MI) Hi-Nic SiC/SiC composite. Time to failure, tf [s] 100 101 102 103 104 105 106 107 Applied shear stress, τ [MPa] 3 4 5 6 7 8 9 20 30 40 10 ns = 22 Hi-Nic SiC/SiC DNS at 1316o C 1min 1h 10h 100h 1000h Fig. 4. Results of stress rupture testing for Hi-Nic SiC/SiC composite in interlaminar shear at 1316 C in air. The solid line represents the best fit. The arrowed data points denote the run out specimens. 892 S.R. Choi et al. / Composites Science and Technology 69 (2009) 890–897
S.R. Chof et aL/ Composites Science and Technology 69(2009)890-897 oken fibers Broken Rmatrices 100m ind region 200pm tows 5. Fracture surfaces showing (a)typical matrix-rich shear region with many fber imprints. (b) tows subjected to shear, and (c)'blind(or cavity regions between two nt tows. The single arrow in each figure represents a direction of shear loading. tical(weft or fill)fiber tows, due to shear, was observed: evidence woven CMCs as well)never revealed such features plainly. This is that the shear load-bearing capacity of the warp or weft fiber tows indeed a daunting challenge in fractography for CMCs as to their was lower than that of the matrix-rich interfaces, thus resulting in nature, origin, and degree of slow crack growth. shear in the tow regions rather than in the matrix-rich regions. It was also found that there were some 'blind regions where no matrix 3. 2. Constant stress-rate tests or silicon was infiltrated between plies or tows, resulting in signifi cant gaps or voids or pores in the composite. This certainly contrib- All specimens tested in constant stress-rate loading failed in utes to decrease in interlaminar shear (and tensile as well) interlaminar shear, similar in mode to those in stress rupture properties of the composite. The common features of fracture sur-(Fig. 3). Most specimens exhibited a linearity between force and faces aforementioned are revealed in Fig. 5. displacement, although the specimen tested at the lowest test rate Fig 6 presents typical fracture surfaces of the specimens tested of 0.0005 MPa/s exhibited some minor nonlinearity because of at a low(10 MPa)and a high(17.8 MPa)applied stresses, where creep deformation involved. The results of constant stress-rate matrix-rich regions, fiber tows shear failure, and 'blind'regions testing are presented in Fig. 7, where failure stress(i.e, interlay are seen in common. In general, the specimens tested at lower nar shear strength) was plotted as a function of applied shear stresses showed somewhat smoother and cleaner fracture surfaces stress rate in a log-log scheme. The dotted line represents the best vith less breakage of fiber tows than those tested at higher stres- fit and the solid line denotes a prediction which will be discussed ses, as shown in the figure. This implies that the matrix-rich inter- in a later section. Despite some scatter in the data, the overall facial failure through slow crack growth or damage accumulation interlaminar shear strength decreased with decreasing applie might have been more predominant at lower stresses than at high- shear stress rates. This phenomenon of strength degradation with er stresses. The time for slow crack growth along the fiber-matrix decreasing test rate often called 'dynamic fatigue when referred interfaces was not sufficient at higher stresses, resulting in a phe- to monolithic brittle materials in tension or in flexure [15, 16] is omenon reminiscent of 'fast fracture that accompanies rough an evidence of slow crack growth or damage accumulation occur- fracture surfaces with increased fiber tow damage. However, it ring at the fiber-matrix interfaces along a respective shear plane should be noted that unlike many homogeneous brittle solids This type of life limiting behavior, associated with strength degra where the regions of slow crack growth are commonly discernable dation in interlaminar shear, was also observed for other CMcs and demarcated from their fracture surfaces, the Cmc(or any other including SiC/SiCs, SiC/MAS, C/SiC, and Sic/BSAS at elevated tem- peratures [6, 7]. Based on both of the results in Figs. 4 and 7, it 2 ar failue occurred excusively at the matri- ich Pegion The ive baess in interlaminar shear, either in static loading(stress rupture)or fabric (e.g, the Hi-Nic Sic siC) was four times longer in warp tow floats time-varying loading( constant stress ale s(stress rupture)or The fracture surfaces of specimens subjected to constant ed to size effect, would be much greater in tows than in matrix- rich re ent Hi-NiC SiC/SiCs exhibited stress-rate tests, in general, were rougher than those of the spec in layers, fi- mpared to that(42+ MPa)of the plain-woven Sic/sic(90 vintage)[. The same ber tows, and matrices. This could be due to the fact that the as true to the case of interlaminar tension strength [5]. extension of slow crack growth was less significant in constant
tical (weft or fill) fiber tows, due to shear, was observed: evidence that the shear load-bearing capacity of the warp or weft fiber tows was lower than that of the matrix-rich interfaces, thus resulting in shear in the tow regions rather than in the matrix-rich regions.1 It was also found that there were some ‘blind’ regions where no matrix or silicon was infiltrated between plies or tows, resulting in signifi- cant gaps or voids or pores in the composite. This certainly contributes to decrease in interlaminar shear (and tensile as well) properties of the composite. The common features of fracture surfaces aforementioned are revealed in Fig. 5. Fig. 6 presents typical fracture surfaces of the specimens tested at a low (10 MPa) and a high (17.8 MPa) applied stresses, where matrix-rich regions, fiber tows shear failure, and ‘blind’ regions are seen in common. In general, the specimens tested at lower stresses showed somewhat smoother and cleaner fracture surfaces with less breakage of fiber tows than those tested at higher stresses, as shown in the figure. This implies that the matrix-rich interfacial failure through slow crack growth or damage accumulation might have been more predominant at lower stresses than at higher stresses. The time for slow crack growth along the fiber–matrix interfaces was not sufficient at higher stresses, resulting in a phenomenon reminiscent of ‘fast fracture’ that accompanies rough fracture surfaces with increased fiber tow damage. However, it should be noted that unlike many homogeneous brittle solids where the regions of slow crack growth are commonly discernable and demarcated from their fracture surfaces, the CMC (or any other woven CMCs as well) never revealed such features plainly. This is indeed a daunting challenge in fractography for CMCs as to their nature, origin, and degree of slow crack growth. 3.2. Constant stress-rate tests All specimens tested in constant stress-rate loading failed in interlaminar shear, similar in mode to those in stress rupture (Fig. 3). Most specimens exhibited a linearity between force and displacement, although the specimen tested at the lowest test rate of 0.0005 MPa/s exhibited some minor nonlinearity because of creep deformation involved. The results of constant stress-rate testing are presented in Fig. 7, where failure stress (i.e., interlaminar shear strength) was plotted as a function of applied shear stress rate in a log–log scheme. The dotted line represents the best fit and the solid line denotes a prediction which will be discussed in a later section. Despite some scatter in the data, the overall interlaminar shear strength decreased with decreasing applied shear stress rates. This phenomenon of strength degradation with decreasing test rate, often called ‘dynamic fatigue’ when referred to monolithic brittle materials in tension or in flexure [15,16], is an evidence of slow crack growth or damage accumulation occurring at the fiber–matrix interfaces along a respective shear plane. This type of life limiting behavior, associated with strength degradation in interlaminar shear, was also observed for other CMCs including SiC/SiCs, SiC/MAS, C/SiC, and SiC/BSAS at elevated temperatures [6,7]. Based on both of the results in Figs. 4 and 7, it can be stated that life limiting behavior of the composite occurred in interlaminar shear, either in static loading (stress rupture) or in time-varying loading (constant stress rate). The fracture surfaces of specimens subjected to constant stress-rate tests, in general, were rougher than those of the specimens in stress rupture, with more damage involved in layers, fi- ber tows, and matrices. This could be due to the fact that the extension of slow crack growth was less significant in constant 1 This was in a marked contrast to the case of the 2D plain-woven SiC/SiC [5], where shear failure occurred exclusively at the matrix-rich region. The five harness satin (HS) fabric (e.g., the Hi-NiC SiC/SiC) was four times longer in warp tow floats than the plain-woven fabric, so that a probability of failure in shear (or in tension), attributed to size effect, would be much greater in tows than in matrix-rich region. Both the state-of-the-art (5HS) MI Sylramic and the current Hi-NiC SiC/SiCs exhibited much lower interlaminar shear strength of 30 ± 5 and 36 ± 3 MPa, respectively, compared to that (42 ± 4 MPa) of the plain-woven SiC/SiC (’90 vintage) [5]. The same was true to the case of interlaminar tension strength [5]. Fig. 5. Fracture surfaces showing (a) typical matrix-rich shear region with many fiber imprints, (b) tows subjected to shear, and (c) ‘blind’ (or cavity) regions between two adjacent tows. The single arrow in each figure represents a direction of shear loading. S.R. Choi et al. / Composites Science and Technology 69 (2009) 890–897 893
SR Choi et al/ Composites Science and Technology 69(2009)890-897 b 1 mm Fig. 6. Fracture surfaces of specimens subjected to stress rupture in interlaminar shear at 1316C in air for Hi-Nic SiC/ siC at: (a)a low st 10 MPa(failure me=17, 246 s)and (b)a high stress of 17.8 MPa(failure time 400 s) Note fiber tow breakage and blind regions. The arrow indicates a direction of shear loading in many homogeneous brittle materials. Typical fracture surfaces Hi-Nic sic/Sic of specimens tested at a high(5 MPa/s)and a low(0.005 MPa/s) G 50FDNS at 1316c stress rates are seen in Fig 8, where fiber imprints, shear breakage of tows, and traces of broken-away tows are seen together with Best fit (n 24) 4. Discussion 4.1. Assessment of life limiting in stress rupture 08765 A phenomenological slow crack growth(SCG)model propose viously [6, 7 will be applied be validated with the constant stress-rate test data. the SCG model in mode Il is similar in expression to the power-law 104 elation in mode I loading, and takes a following, empirical formu ig. 7. Results of constant stress-rate testing for Hi-Nic SiC/SiC composite in \9 Applied shear stress rate, T [MPa/s nterlaminar shear at 1316C in air. The dotted line represents the best fit. prediction from the stress rupture data was also included. where I, a, t, Ki, and Klc are crack growth rate in interlaminar shear, crack size, time mode ll stress intensity factor, and mode ll fracture toughness, respectively. as and ns are life limiting(or stress-rate tests than in stress rupture, because of shortened test scG) parameters in interlaminar shear. The above formulation times in constant stress rate tests. The fracture surfaces of speci- as fracture mechanics based, in which a crack residing at ma- ens at lower stress rates were somewhat smoother lose of trix-rich interfacial regions is assumed to grow subcritically, even- pecimens tested at higher test rates, presumably an indication of tually leading to an instability condition. It may be arguable as to more extended slow crack growth occurring at lower stress rates. whether a term 'crack or fracture mechanics concept can be used The enhanced degree of slow crack growth at lower test rates for CMCs since fracture origin and subsequent crack growth region might have yielded smoother fracture surfaces, as also observed are not typically observable from fracture surfaces, due to their b 1mm Fig. 8. Fracture surfaces of specimens subjected to constant stress-rate testing in interlaminar shear at 1316C in air for Hi-Nic Sic/ SiC composite at: (a)a low stress rate of 0.005 MPa/s and (b)a high stress rate of 5 MPa s Note fiber tow breakage and"blind regions. The arrow indicates a direction of shear loading
stress-rate tests than in stress rupture, because of shortened test times in constant stress rate tests. The fracture surfaces of specimens at lower stress rates were somewhat smoother than those of specimens tested at higher test rates, presumably an indication of more extended slow crack growth occurring at lower stress rates. The enhanced degree of slow crack growth at lower test rates might have yielded smoother fracture surfaces, as also observed in many homogeneous brittle materials. Typical fracture surfaces of specimens tested at a high (5 MPa/s) and a low (0.005 MPa/s) stress rates are seen in Fig. 8, where fiber imprints, shear breakage of tows, and traces of broken-away tows are seen together with ‘blank’ regions. 4. Discussion 4.1. Assessment of life limiting in stress rupture A phenomenological slow crack growth (SCG) model proposed previously [6,7] will be applied to the stress rupture data and will be validated with the constant stress-rate test data. The proposed SCG model in mode II is similar in expression to the power-law relation in mode I loading, and takes a following, empirical formulation [6,7]: vs ¼ da dt ¼ asðKII=KIIcÞ ns ð3Þ where vs, a, t, KII, and KIIc are crack growth rate in interlaminar shear, crack size, time, mode II stress intensity factor, and mode II fracture toughness, respectively. as and ns are life limiting (or SCG) parameters in interlaminar shear. The above formulation was fracture mechanics based, in which a crack residing at matrix-rich interfacial regions is assumed to grow subcritically, eventually leading to an instability condition. It may be arguable as to whether a term ‘crack’ or fracture mechanics concept can be used for CMCs since fracture origin and subsequent crack growth region are not typically observable from fracture surfaces, due to their Fig. 6. Fracture surfaces of specimens subjected to stress rupture in interlaminar shear at 1316 C in air for Hi-Nic SiC/SiC composite at: (a) a low stress of 10 MPa (failure time = 17,246 s) and (b) a high stress of 17.8 MPa (failure time = 400 s). Note fiber tow breakage and ‘blind’ regions. The arrow indicates a direction of shear loading. Applied shear stress rate, [MPa/s] 10-4 10-3 10-2 10-1 100 101 102 Failure stress, τf [MPa] 5 6 7 8 20 30 40 50 60 70 10 Hi-Nic SiC/SiC DNS at 1316o C Best fit (ns=24) Prediction (ns=22) τ Fig. 7. Results of constant stress-rate testing for Hi-Nic SiC/SiC composite in interlaminar shear at 1316 C in air. The dotted line represents the best fit. A prediction from the stress rupture data was also included. Fig. 8. Fracture surfaces of specimens subjected to constant stress-rate testing in interlaminar shear at 1316 C in air for Hi-Nic SiC/SiC composite at: (a) a low stress rate of 0.005 MPa/s and (b) a high stress rate of 5 MPa/s. Note fiber tow breakage and ‘blind’ regions. The arrow indicates a direction of shear loading. 894 S.R. Choi et al. / Composites Science and Technology 69 (2009) 890–897
S.R. Chof et al / Composites Science and Technology 69(2009)890-897 rchitectural complexity, as aforementioned. So, in principle, any 4.2. Verification other approaches such as creep or damage mechanics, if appropri ate, can be utilized as an alternative to lifing. The proposed SCG formulation, Eq. (3), indicates that for a given he generalized expression of Ki along the crack front of a pen- material/environmental condition, crack velocity depends on Ku So ny-or half-penny shaped crack subjected to shear loading either that in principle the life limiting parameters can be determined in on crack planes or on remote material body can take the following any loading configuration which is either static, cyclic, or any time form [17: varying. Therefore, it should be feasible to make a life prediction Ku=Ysta/f(0, failure mechanism is operative. In this section, the life limiting where Ys is a crack geometry factor related to a function of fe, o) parameters that were determined from stress rupture will be used with the angles 0 and o being related to load and a particular point to predict the strength degradation behavior in constant stress-rate of the crack front. t is the applied interlaminar shear stress. Using loading and to validate the proposed SCG model. Note again that Eqs. (3)and (4)together with some mathematical manipulations, strength degradation with respect to decreasing test rate is another one can obtain time to failure(ta) as a function of applied shear form of life limiting phenomena, where the degree of strength deg stress, as done for brittle materials in mode I loading [18 adation actually represents the extent of SCG [15, 16. Using eqs (5)and(6) with some mathematical manipulations, interlaminar t= Dst (5) shear strength(tr) as a function of applied shear stress rate(t) can be derived as follows. D,=B[tj s- tr=Dd可 where B,=2Knc,y3(ns-2). t, is the inert shear strength that is defined as a strength with no SCG, and the geometry function is Da=D(ns +1)s+T lified as fe, )=1 in the case of double notch shear loading for an infinite material body. Eq (5)can be expressed in a more Eq.(9)indicates that for a given material environment inter onvenient form by taking logarithms of both sides laminar shear strength in constant stress-rate loading can be pre- dicted as a function of stress rate once ns and ds from stress log t=-ns log t + log D 7 rupture data are known. The resulting prediction of interlaminar shear strength, based monolithic ceramics[18, 19). Life limiting parameters ns and Ds in Fig. i. Despite some scatters in shear strength, the prediction interlaminar shear can be determined based on Eg (7), respectively. from the slope and the intercept of a linear regression analysis of was in good agreement with the best fit of the experimental data the log(individual applied interlaminar shear stress with units The discrepancy in shear strength between the prediction and the data was 6-9%. Particularly, the life limiting parameter of ns =22 MPa)vs log(individual time to failure with unit of second)data, estimated from stress rupture was in excellent agreement with Fig 4, which results in ns=24t8, determined from the constant stress-rate data by a n3=22±5;D5=28±8 ession analysis of log(to)-vS -log(t), Eq.(8). The predicted with the coefficient of correlation of curve fit of rcoef=0.7303. The parameter log D=19 from Eq (9)was also in good agreement with regression fit was indicated as a solid line in the figure. As seen from results showed that the overal the constant stress-rate data.These g Da=21 t 1 estimated from the figure, statistically good agreement exists between the model yearning failure mechanism of the few orders of magnitude, which is typical of most brittle materials almost unchanged, regardless of loading configurations and that in stress rupture or in cyclic fatigue. The parameter as in Eq (3)can of SCG formulation, Eq(3). Statistically, the prediction made above be evaluated using the Bs relation together with appropriate param- eters/constants. represents a failure probability of approximately 50% The empirical SCG formulation of homogeneous brittle solids 4.3. Statistical aspects of the dato subjected to mode I loading is expressedas da/dt=x(K/K)with K KIc, a, and n being stress intensity factor,fracture toughness, and SCG parameters, respectively. It life/strength predictions was indicative of a seemingly same failure has been generally categorized for brittle materials under mode mechanism prevailing in the composite regardless of loading con- figurations. This prompts to see if statistical aspects of the data ate for n- 30-40, and insignificant for n> 50 [6]. Hence, based would follow a pattern similar to that of brittle materials in mode can be said to exhibit a significant life susceptibility in interlaminar L. It has been shown that under mode I loading, brittle materials on this categorization, the Hi-Nic SiC/Sic composite with ns =22 shear at 1316C in air. However, it should be noted that the use of when exhibit one dominant failure mechanism, yield a relationship the categorization was not based on any physical correlation be. in (2-parameter) Weibull modulus between strength and time to tween n and ns. values of n and ns in various CMCs have been com- failure as follows [19] pared and reported previously [10]. where ms and md are Weibull modulus in time to failure distribu- tions(in stress rupture)and Weibull modulus in strength distribu- factor Y(e,)can change as a crack grows through a SCG process if tions(in constants stress rate), respectively meter n > 20, resulting in little change in the value of time to nstant stress-rate loading)[20). Hence, the approach using fe, o)=1 with a 3 The relationship can be derived from Eqs. (5)and(9)if Weibull distribution of tant Ys in this work could be reasonable throughout the lives of test coupons. inert shear strength(t) is given
architectural complexity, as aforementioned. So, in principle, any other approaches such as creep or damage mechanics, if appropriate, can be utilized as an alternative to lifing. The generalized expression of KII along the crack front of a penny- or half-penny shaped crack subjected to shear loading either on crack planes or on remote material body can take the following form [17]: KII ¼ Yssa1=2fðh;uÞ ð4Þ where Ys is a crack geometry factor related to a function of f(h,u) with the angles h and u being related to load and a particular point of the crack front. s is the applied interlaminar shear stress. Using Eqs. (3) and (4) together with some mathematical manipulations, one can obtain time to failure (tf) as a function of applied shear stress, as done for brittle materials in mode I loading [18] tf ¼ Ds½s ns ð5Þ where Ds ¼ Bs½si ns2 ð6Þ where Bs ¼ 2KIIc asY2 s ðns 2Þ h i, si is the inert shear strength that is defined as a strength with no SCG, and the geometry function is simplified as f(h,u) = 1 in the case of double notch shear loading for an infinite material body.2 Eq. (5) can be expressed in a more convenient form by taking logarithms of both sides log tf ¼ ns log s þ log Ds ð7Þ which is identical in form to the case in mode I loading used in monolithic ceramics [18,19]. Life limiting parameters ns and Ds in interlaminar shear can be determined based on Eq. (7), respectively, from the slope and the intercept of a linear regression analysis of the log (individual applied interlaminar shear stress with units of MPa) vs. log (individual time to failure with unit of second) data, Fig. 4, which results in ns ¼ 22 5; Ds ¼ 28 8 with the coefficient of correlation of curve fit of rcoef = 0.7303. The regression fit was indicated as a solid line in the figure. As seen from the figure, statistically good agreement exists between the model and the data, although the data scatter in time to failure was in a few orders of magnitude, which is typical of most brittle materials in stress rupture or in cyclic fatigue. The parameter as in Eq. (3) can be evaluated using the Bs relation together with appropriate parameters/constants. The empirical SCG formulation of homogeneous brittle solids subjected to mode I loading is expressed as v ¼ da=dt ¼ aðKI=KIc Þ n with KI, KIc, a, and n being stress intensity factor, fracture toughness, and SCG parameters, respectively. It has been generally categorized for brittle materials under mode I that life limiting susceptibility is significant for n < 30, intermediate for n 30–40, and insignificant for n P 50 [6]. Hence, based on this categorization, the Hi-Nic SiC/SiC composite with ns = 22 can be said to exhibit a significant life susceptibility in interlaminar shear at 1316 C in air. However, it should be noted that the use of the categorization was not based on any physical correlation between n and ns. Values of n and ns in various CMCs have been compared and reported previously [10]. 4.2. Verification The proposed SCG formulation, Eq. (3), indicates that for a given material/environmental condition, crack velocity depends on KII so that in principle the life limiting parameters can be determined in any loading configuration which is either static, cyclic, or any time varying. Therefore, it should be feasible to make a life prediction from one loading configuration to another provided that the same failure mechanism is operative. In this section, the life limiting parameters that were determined from stress rupture will be used to predict the strength degradation behavior in constant stress-rate loading and to validate the proposed SCG model. Note again that strength degradation with respect to decreasing test rate is another form of life limiting phenomena, where the degree of strength degradation actually represents the extent of SCG [15,16]. Using Eqs. (5) and (6) with some mathematical manipulations, interlaminar shear strength (sf) as a function of applied shear stress rate ðs_Þ can be derived as follows: sf ¼ Dd½s_ 1 nsþ1 ð8Þ where Dd ¼ ½Dsðns þ 1Þ 1 nsþ1 ð9Þ Eq. (9) indicates that for a given/material environment interlaminar shear strength in constant stress-rate loading can be predicted as a function of stress rate once ns and Ds from stress rupture data are known. The resulting prediction of interlaminar shear strength, based on Eqs. (8) and (9) with ns and Ds, is presented as a solid line in Fig. 7. Despite some scatters in shear strength, the prediction was in good agreement with the best fit of the experimental data. The discrepancy in shear strength between the prediction and the data was 6–9%. Particularly, the life limiting parameter of ns = 22 estimated from stress rupture was in excellent agreement with ns = 24 ± 8, determined from the constant stress-rate data by a regression analysis of log (sf)-vs.-logðs_Þ, Eq. (8). The predicted parameter logDd = 19 from Eq. (9) was also in good agreement with logDd = 21 ± 1 estimated from the constant stress-rate data. These results showed that the overall governing failure mechanism of the composite subjected to interlaminar shear might have remained almost unchanged, regardless of loading configurations and that the failure mechanism could be described by the power-law type of SCG formulation, Eq. (3). Statistically, the prediction made above represents a failure probability of approximately 50%. 4.3. Statistical aspects of the data As stated in the preceding section, the good agreement in the life/strength predictions was indicative of a seemingly same failure mechanism prevailing in the composite regardless of loading con- figurations. This prompts to see if statistical aspects of the data would follow a pattern similar to that of brittle materials in mode I. It has been shown that under mode I loading, brittle materials, when exhibit one dominant failure mechanism, yield a relationship in (2-parameter) Weibull modulus between strength and time to failure as follows [19]3 ms md n þ 1 ð10Þ where ms and md are Weibull modulus in time to failure distributions (in stress rupture) and Weibull modulus in strength distributions (in constants stress rate), respectively. 2 The geometry factor Ys f(h,u) can change as a crack grows through a SCG process if the crack becomes finite relative to the material body (test coupon). This change, however, occurs typically close to an instability for a material exhibiting a life limiting parameter n P 20, resulting in little change in the value of time to failure (or strength in constant stress-rate loading) [20]. Hence, the approach using f(h,u) = 1 with a constant Ys in this work could be reasonable throughout the lives of test coupons. 3 The relationship can be derived from Eqs. (5) and (9) if Weibull distribution of inert shear strength (si) is given. S.R. Choi et al. / Composites Science and Technology 69 (2009) 890–897 895
896 SR Choi et al/ Composites Science and Technology 69(2009)890-897 10 40 MPa 1x1021x10 10° 1x1 Hi-Nic SiC/SiC ILS/1316.C Hi-Nic SIC/SIC ILS/1316C 5 MPa/s 0.158MPa/s ▲t=12.6MPa ▲0.005MPa/s v t=10.0 MPa =6.5 2 3 5 In( )[MPa In(t, )[sec (a) Constant stress-rate (b)Stress rupture Fig 9. Weibull distributions of shear strength and time to failure for Hi-Nic SiC/SiC composite in interlaminar shear tested at 1316Cin air (a)Shear strength distributions in ant stress-rate testing: The solid lines represent the functional fits with md=6.5, averaged over three individual Weibull moduli in constant stress-rate tests and(b)time to failure distributions in stress rupture: The solid lines represent the functional fits based on ms =0.3, predicted from Eq (10) with md=6.5. Fig. 9 represents the 2-parameter Weibull plots and related which is time consuming particularly at lower applied stresses. functional fits for shear strength and time to failure. Weibull mod- By contrast, constant stress-rate testing gives merits in simplicity, ulus in interlaminar shear strength of the composite was md=5.5. test economy (short test times), and less data scatter thus allowing 6.1, and 8.0 at t=5.0.158 and 0.005 MPa/s, respectively, resulting less number of test specimens. However, caution should be exer in an average of md=6.5+1.3. Use of Eq (10) with md= 6.5 and cised when more than one failure mechanism occur, combined n=ns=22 yielded a predicted value of ms=0.3. The individual with SCG, fatigue, creep and environmental degradation, etc In this Weibull modulus in time to failure was found to be m,=0.7, 0.4, case, use of constant stress-rate testing should be limited for the 0.4, and 0.3 at t=17.8, 15.0, 12.6, and 10 MPa, respectively, with case of a short period of lives. a resulting average of ms=0. 4+0. 2. Therefore, the predicted Exploration of detailed microscopic failure mechanism(s)associ ms=0.3 based on Eq (10)was in reasonable agreement with the ated in interlaminar shear is beyond the scope of this work. More actual ms=0. 4, which is also seen from Fig. 9b. Considering the detailed analyses involved with matrix/fiber interaction, matrix microstructure and architecture, behaves as if it were a homoge- enological model proposed here may incorporate other operative neous material in mode l. These statistical aspects of the composite models such as viscous sliding, void nucleation, growth and coales- draw again an important conclusion that the governing failure cence, etc, which can be all covered under a generic term of delayed mechanism would have remained almost invariant either in stress failure, slow crack growth, fatigue, creep, or damage initiation/ a rupture or in constant stress rate. Note that the relation in Eq (10) mulation. Furthermore additional tests over a wide range of tem- was derived with a very basic assumption that failure mechanism peratures would be necessary to identify in more detail the failure remains consistent, independent of the type of loading [19]. Cer- mechanisms since activation energy could then be established and tainly, use of more test specimens could better verify the validity a temperature-compensated time method could be used to help of the relation of Eq (10) for the composite. One of the possible fit experimental data with an increased accuracy [24. ources of the data variability may have been from the notch geometry, particularly the location of notch tip relative to the mid-plane of test specimens, since the surfaces of test specimens 5 Conclusions used as a datum were not machined but as-processed. The material and processing themselves, of course, might have been other major Life limiting behavior of Hi-Nic SiC/SiC composite was assessed sources of the variability as well. in interlaminar shear using stress rup ting at1316° in air The composite exhibited a significant life limiting susceptibility 4.4. Life prediction methodology of CMCs in interlaminar shear with ns =22 which was evaluated based on a proposed phenom nological crack growth law. Good agreement was found in inter- The results of this work indicate that the governing failure laminar shear strength between the prediction from the stress mechanism in interlaminar shear would not differ significantly rupture data and the actual constant stress-rate data Statistica either in stress rupture or in constant stress-rate loading and that pects of Weibull modulus, which have been found in mode I for the overall failure mechanism could be described by the power-law many homogeneous brittle materials, existed between the time type of SCG formulation, Eq.(3). The results also indicate that to failure and the strength data. The results indicated that the gov- stress rupture or constant stress-rate testing can be utilized as a erning failure mechanism(s) would have remained likely un- life prediction test methodology to determine the phenomenolog changed either in stress rupture or in constant stress rate ical life prediction parameters of CMCs even in interlaminar shear. loading. Constant stress-rate testing could be a possible means of Stress rupture testing gives more realistic life data but it gives in- life prediction test methodology for CMCs in interlaminar shear creased data scatter so that more test specimens are required, when short lifetimes of components are anticipated
Fig. 9 represents the 2-parameter Weibull plots and related functional fits for shear strength and time to failure. Weibull modulus in interlaminar shear strength of the composite was md = 5.5, 6.1, and 8.0 at s_ ¼ 5; 0:158 and 0:005MPa=s, respectively, resulting in an average of md = 6.5 ± 1.3. Use of Eq. (10) with md = 6.5 and n ns = 22 yielded a predicted value of ms = 0.3. The individual Weibull modulus in time to failure was found to be ms = 0.7, 0.4, 0.4, and 0.3 at s = 17.8, 15.0, 12.6, and 10 MPa, respectively, with a resulting average of ms = 0.4 ± 0.2. Therefore, the predicted ms = 0.3 based on Eq. (10) was in reasonable agreement with the actual ms = 0.4, which is also seen from Fig. 9b. Considering the limited number of test specimens used (typically five), this agreement was surprising since the composite, which is so complex in microstructure and architecture, behaves as if it were a homogeneous material in mode I. These statistical aspects of the composite draw again an important conclusion that the governing failure mechanism would have remained almost invariant either in stress rupture or in constant stress rate. Note that the relation in Eq. (10) was derived with a very basic assumption that failure mechanism remains consistent, independent of the type of loading [19]. Certainly, use of more test specimens could better verify the validity of the relation of Eq. (10) for the composite. One of the possible sources of the data variability may have been from the notch geometry, particularly the location of notch tip relative to the mid-plane of test specimens, since the surfaces of test specimens used as a datum were not machined but as-processed. The material and processing themselves, of course, might have been other major sources of the variability as well. 4.4. Life prediction methodology of CMCs in interlaminar shear The results of this work indicate that the governing failure mechanism in interlaminar shear would not differ significantly either in stress rupture or in constant stress-rate loading and that the overall failure mechanism could be described by the power-law type of SCG formulation, Eq. (3). The results also indicate that stress rupture or constant stress-rate testing can be utilized as a life prediction test methodology to determine the phenomenological life prediction parameters of CMCs even in interlaminar shear. Stress rupture testing gives more realistic life data but it gives increased data scatter so that more test specimens are required, which is time consuming particularly at lower applied stresses. By contrast, constant stress-rate testing gives merits in simplicity, test economy (short test times), and less data scatter thus allowing less number of test specimens. However, caution should be exercised when more than one failure mechanism occur, combined with SCG, fatigue, creep and environmental degradation, etc. In this case, use of constant stress-rate testing should be limited for the case of a short period of lives. Exploration of detailed microscopic failure mechanism(s) associated in interlaminar shear is beyond the scope of this work. More detailed analyses involved with matrix/fiber interaction, matrix and fiber cracking, localized SCG, creep, and environmental effects [21–23] need to be done. It should be kept in mind that the phenomenological model proposed here may incorporate other operative models such as viscous sliding, void nucleation, growth and coalescence, etc., which can be all covered under a generic term of delayed failure, slow crack growth, fatigue, creep, or damage initiation/accumulation. Furthermore, additional tests over a wide range of temperatures would be necessary to identify in more detail the failure mechanisms since activation energy could then be established and a temperature-compensated time method could be used to help fit experimental data with an increased accuracy [24]. 5. Conclusions Life limiting behavior of Hi-Nic SiC/SiC composite was assessed in interlaminar shear using stress rupture testing at 1316 C in air. The composite exhibited a significant life limiting susceptibility with ns = 22 which was evaluated based on a proposed phenomenological crack growth law. Good agreement was found in interlaminar shear strength between the prediction from the stress rupture data and the actual constant stress-rate data. Statistical aspects of Weibull modulus, which have been found in mode I for many homogeneous brittle materials, existed between the time to failure and the strength data. The results indicated that the governing failure mechanism(s) would have remained likely unchanged either in stress rupture or in constant stress rate loading. Constant stress-rate testing could be a possible means of life prediction test methodology for CMCs in interlaminar shear when short lifetimes of components are anticipated. ln(τ f ) [MPa] 2.0 2.5 3.0 3.5 4.0 lnln(1/(1-F)) -3 -2 -1 0 1 2 5 MPa/s 0.158 MPa/s 0.005 MPa/s Hi-Nic SiC/SiC ILS/1316o C md =6.5 10 20 30 40 [MPa] ln(tf ) [sec] -5 0 5 10 15 20 lnln(1/(1-F)) -3 -2 -1 0 1 2 τ =17.8 MPa τ =12.6 MPa τ =10.0 MPa τ =15.0 MPa Hi-Nic SiC/SiC ILS/1316o C ms=0.3 1 1x102 1x104 1x106 1x108 [sec] (a) Constant stress-rate (b) Stress rupture Fig. 9. Weibull distributions of shear strength and time to failure for Hi-Nic SiC/SiC composite in interlaminar shear tested at 1316 C in air. (a) Shear strength distributions in constant stress-rate testing: The solid lines represent the functional fits with md = 6.5, averaged over three individual Weibull moduli in constant stress-rate tests and (b) time to failure distributions in stress rupture: The solid lines represent the functional fits based on ms = 0.3, predicted from Eq. (10) with md = 6.5. 896 S.R. Choi et al. / Composites Science and Technology 69 (2009) 890–897
S.R. Chot et al/Composites Science and Technology 69(2009)890-897 Acknowledg [12] ASTM C 1259. Annual book of ASTM sta Test method for dynamic Youngs modulus, shear modulus, and po The authors(navair)acknowledge Dr D shifler of Naval Research for the support of this work. Some mechanical [13] Choi SR, Kowalik Rw Interlaminar crack growth resistances of various ceramic testing was conducted at the NAsa glenn, Cleveland I and mode ll loading. J Eng Gas Turb Power [14] ASTM C 1425. Test method References voL. 15.01. West Conshohocken(PA): Annual book of ASTM [1 Brondsted P, Heredia PE, Evans AG In-plane shear properties of 2-D ceramic [15] ASTM C 1368. Standard test method for [2] Lara-Curzio E, Ferber MK. Shear strength of continuous fiber ceramic mposites ASTM STP 1309. West Conshohocken(PA): American Society for st Conshohocken (PA): American Society for Testing and Materials: [3 Fang NI. Chou Tw. Characterization of interlaminar shear strength of ceramic [16] ASTM C 1465. Standard test method for determination of slow advanced ceramics by constant stress-rate flexural testing at [5] Choi SR, Bansal NP. Interlaminar tension/shear properties and stress rupture in [171 Tada H, Paris PC, Irwin GR The stress analysis of cracks handbook. NY:ASME eram trans2006;175:119-34 18] ASTM C1576.5 [6 Choi SR, Bansal NP. Shear strength as a function of test rate for SiC/BSAS amic matrix composite at elevated temperature. J Am Ceram Soc West Conshohocken(PA) n Society for Testing and Materials; [71 Choi SR, Bansal NP, Calomino AM, Verrilli M). Shear strength behavior of [19] Ritter JE, Bandyopadhyay N, Jakus k statistical reproducibility of the dynamic 2005:165:131-45 [20] Choi SR, Gyekenyesi JP. Slow crack growth analysis of brittle materials with MB. Effects of temperature and steam environment subjected to constant stress-rate testing. J Mater Sci creep behavior ide-oxide ceramic composite. Ceram Eng Sci Pro 199934:38 2 CA, Henager CH, Jones RH. Environmentally induced time- Alexander DJ Bansal NP. Assessments of life limiting oc2007;28(2):179-89 22]H cal crack growth in Cvi silicon carbide [10 IDEM. Elevated-temperature life limiting behavior of Hi-Nic SiC/Sic ceramic vith Nicalon experiment and model. J Am Ceram Soc 1994:77(9):2381-94. of the Navy. Naval Air Warfare Center Aircraft Division, Patuxent River. MD: [231 Spearing SM. Zok Fw. Evans AG. Stress co 562-70. [111 Brewer D. HSR/EPM combustor materials development program. Mater Sci Eng [24 DiCalo JA Creep and rupture behavior of advanced Sic fibers. In: Proc. ICCM-
Acknowledgements The authors (NAVAIR) acknowledge Dr. D. Shifler and the Office of Naval Research for the support of this work. Some mechanical testing was conducted at the NASA Glenn, Cleveland, Ohio. References [1] Brondsted P, Heredia PE, Evans AG. In-plane shear properties of 2-D ceramic composites. J Am Ceram Soc 1994;77(10):2569–74. [2] Lara-Curzio E, Ferber MK. Shear strength of continuous fiber ceramic composites. ASTM STP 1309. West Conshohocken (PA): American Society for Testing and Materials; 1997. p. 31. [3] Fang NJJ, Chou TW. Characterization of interlaminar shear strength of ceramic matrix composites. J Am Ceram Soc 1993;76(10):2539–48. [4] Ünal Ö, Bansal NP. In-plane and interlaminar shear strength of a unidirectional Hi-Nicalon fiber-reinforced celsian matrix composite. Ceram Int 2002;28:527–40. [5] Choi SR, Bansal NP. Interlaminar tension/shear properties and stress rupture in shear of various continuous fiber-reinforced ceramic matrix composites. Ceram Trans 2006;175:119–34. [6] Choi SR, Bansal NP. Shear strength as a function of test rate for SiCf/BSAS ceramic matrix composite at elevated temperature. J Am Ceram Soc 2004;87(10):1912–8. [7] Choi SR, Bansal NP, Calomino AM, Verrilli MJ. Shear strength behavior of ceramic matrix composites at elevated temperatures. Ceram Trans 2005;165:131–45. [8] Braun JC, Ruggles-Wrenn MB. Effects of temperature and steam environment on creep behavior of an oxide-oxide ceramic composite. Ceram Eng Sci Pro 2008;29(2):51–66. [9] Choi SR, Kowalik RW, Alexander DJ, Bansal NP. Assessments of life limiting behavior in interlaminar shear for Hi-Nic SiC/SiC ceramic matrix composite at elevated temperature. Ceram Eng Sci Proc 2007;28(2):179–89. [10] IDEM. Elevated-temperature life limiting behavior of Hi-Nic SiC/SiC ceramic matrix composite in interlaminar shear. NAWCADPAX/TR-2007/5, Department of the Navy, Naval Air Warfare Center Aircraft Division, Patuxent River, MD; 2007. [11] Brewer D. HSR/EPM combustor materials development program. Mater Sci Eng 1999;A261:284–91. [12] ASTM C 1259. Annual book of ASTM standards. Test method for dynamic Young’s modulus, shear modulus, and Poisson’s ratio for advanced ceramics by impulse excitation of vibration, vol. 15.01. West Conshohocken (PA): American Society for Testing and Materials; 2006. [13] Choi SR, Kowalik RW. Interlaminar crack growth resistances of various ceramic matrix composites in mode I and mode II loading. J Eng Gas Turb Power 2008;130:1301-1-8. [14] ASTM C 1425. Test method for interlaminar shear strength of 1-D and 2-D continuous fiber-reinforced advanced ceramics at elevated temperatures, vol. 15.01. West Conshohocken (PA): Annual book of ASTM standards; 2006. [15] ASTM C 1368. Standard test method for determination of slow crack growth parameters of advanced ceramics by constant stress-rate flexural testing at ambient temperature. Annual book of ASTM standards, vol. 15.01. West Conshohocken (PA): American Society for Testing and Materials; 2006. [16] ASTM C 1465. Standard test method for determination of slow crack growth parameters of advanced ceramics by constant stress-rate flexural testing at elevated temperatures. Annual book of ASTM standards, vol. 15.01. West Conshohocken (PA): American Society for Testing and Materials; 2006. [17] Tada H, Paris PC, Irwin GR. The stress analysis of cracks handbook. NY: ASME; 2000. p. 418. [18] ASTM C 1576. Standard test method for determination of slow crack growth parameters of advanced ceramics by constant stress flexural testing (stress rupture) at ambient temperature. Annual book of ASTM standards, vol. 15.01. West Conshohocken (PA): American Society for Testing and Materials; 2006. [19] Ritter JE, Bandyopadhyay N, Jakus K. Statistical reproducibility of the dynamic and static fatigue experiments. Ceram Bull 1981;60(8):798–806. [20] Choi SR, Gyekenyesi JP. Slow crack growth analysis of brittle materials with finite thickness subjected to constant stress-rate testing. J Mater Sci 1999;34:3875–82. [21] Lewinsohn CA, Henager CH, Jones RH. Environmentally induced timedependent failure mechanism in CFCCS at elevated temperatures. Ceram Eng Sci Proc 1998;19(4):11–8. [22] Henager CH, Jones RH. Subcritical crack growth in CVI silicon carbide reinforced with Nicalon fibers: experiment and model. J Am Ceram Soc 1994;77(9):2381–94. [23] Spearing SM, Zok FW, Evans AG. Stress corrosion cracking in a unidirectional ceramic–matrix composite. J Am Ceram Soc 1994;77(2):562–70. [24] DiCalo JA. Creep and rupture behavior of advanced SiC fibers. In: Proc. ICCM- 10, vol. 6; 1995. p. 315 (1995). S.R. Choi et al. / Composites Science and Technology 69 (2009) 890–897 897