./. Appl. Ceram. Technol, 612/151-163(2009) DO:10.J11.1747402.2008.02331.x Applied Ceramic Technolog ceramic Product Development and Commercialization Design Guidelines for In-Plane Mechanical Properties of SiC Fiber-Reinforced Melt-Infiltrated SiC Composites Ohio Aerospace Institute, Cleveland, Ohio 44142 Vijay V.Pt Materials and Simulation Technical Center, Goodrich Corporation, Brecksville, Ohio 44141 In-plane tensile stress-strain, tensile creep, and after-creep retained tensile properties of melt- infiltrated SiC-SiC com- posites reinforced with different fiber types were evaluated with an emphasis on obtaining simple or first-order microstructural design guidelines for these in-plane mechanical properties. Using the minimatrix approach to model stress-strain behavior and he results of this study, three basic general design criteria for stress and strain limits are formulated, namely a design stress limit, a design total strain limit, and an after-creep design retained strength limit. It is shown that these criteria can be useful for Introduction for composite designers and fabricators who often have to weigh the benefits of cost savings, for example, a less Woven silicon carbide fiber-reinforced melt- infil- expensive fiber, with performance targets demanded by rated(MD)silicon carbide matrix composites are an application. Constituent-based and fiber-architec- considered to be important enabling materials for ture-based design models need to be developed to as- composites(CMC); however, even within that subset, given application. To validate these models, composite onsiderable variations in thermomechanical properties property data are needed over wide variations in con- are possible depending on the composite constituent stituent compositions, geometries, and content. A con- materials, geometries, and content. This is important siderable amount of data has been generated on composites with a wide variation in fber fraction and fiber architectures for the Sylramic-iBn (Syl-iBN)fiber- ork was financially supported by both internal Goodrich and NASA Supers reinforced MI composite system. 4 ./- These results .grams as well as a partially reimbursable Space Act Agreement between Goodrich and have enabled the development of simple constituent based and fiber-architecture- based relationships that can an Ceramic guide designers, fabricators, and end users in predicting
Design Guidelines for In-Plane Mechanical Properties of SiC Fiber-Reinforced Melt-Infiltrated SiC Composites Gregory N. Morscher* Ohio Aerospace Institute, Cleveland, Ohio 44142 Vijay V. Pujar Materials and Simulation Technical Center, Goodrich Corporation, Brecksville, Ohio 44141 In-plane tensile stress–strain, tensile creep, and after-creep retained tensile properties of melt-infiltrated SiC–SiC composites reinforced with different fiber types were evaluated with an emphasis on obtaining simple or first-order microstructural design guidelines for these in-plane mechanical properties. Using the minimatrix approach to model stress–strain behavior and the results of this study, three basic general design criteria for stress and strain limits are formulated, namely a design stress limit, a design total strain limit, and an after-creep design retained strength limit. It is shown that these criteria can be useful for designing components for high-temperature applications. Introduction Woven silicon carbide fiber-reinforced melt-infiltrated (MI) silicon carbide matrix composites are considered to be important enabling materials for high-temperature turbine1,2 applications. MI matrix SiC composites are a subset of SiC/SiC ceramic matrix composites (CMC); however, even within that subset, considerable variations in thermomechanical properties are possible depending on the composite constituent materials, geometries, and content.3–6 This is important for composite designers and fabricators who often have to weigh the benefits of cost savings, for example, a less expensive fiber, with performance targets demanded by an application. Constituent-based and fiber-architecture-based design models need to be developed to assess whether a given fiber type can meet the property requirements or the cost/performance objectives for a given application. To validate these models, composite property data are needed over wide variations in constituent compositions, geometries, and content. A considerable amount of data has been generated on composites with a wide variation in fiber fraction and fiber architectures for the Sylramic-iBN (Syl-iBN) fiberreinforced MI composite system.4,7–11 These results have enabled the development of simple constituentbased and fiber-architecture-based relationships that can guide designers, fabricators, and end users in predicting Int. J. Appl. Ceram. Technol., 6 [2] 151–163 (2009) DOI:10.1111/j.1744-7402.2008.02331.x Ceramic Product Development and Commercialization This work was financially supported by both internal Goodrich and NASA Supersonic programs as well as a partially reimbursable Space Act Agreement between Goodrich and NASA. *gregory.n.morscher@nasa.gov r 2008 The American Ceramic Society
International Journal of Applied Ceramic Technolog-Morscher and pujar Vol.6,No.2,2009 properties such as matrix cracking stresses, ultimate ten- dustries, Tokyo, Japan). In addition, Table I also in- ile properties, and elevated temperature creep and cludes data from two panels with Hi-Nicalon Type- atigue properties Compared with other commercially (Nippon Carbon) fber that came from the earli far available fibers, the Syl-iBN fber evaluated in these study, 'which is included in this paper for property studies is very stable against high-temperature degrade stud comparison. For convenience, the composite panels are tion both during processing and service, and as a result referred to as xxx-Y where xxx is the reinforcing fiber is expected to be less prone to mechanical performance type(Syl-iBN, SA, HN, ZMI, HNS)and Y is the panel variation arising from process and/or application varia- number with that particular fiber tions. However, the Syl-iBN fiber is not commercially For in-plane mechanical property evaluation ailable readily, and the other fiber types may be more tensile specimens, 150 mm long and 12.6 mm wide attractive as they offer an overall cost advantage over the at the ends, were machined from the panels into a Syl-iBN fiber in meeting the necessary property require- dog-bone shape where the gauge section length and ments for some applications. width were approximately 25 and 10 mm, respectively The purpose of this study was to assess the in-plane The length of each specimen was aligned as close as mechanical performance of 2D 0/90 MI composites possible with one of the two orthogonal fiber directions, (oriented in one of the orthogonal fiber directions) commonly referred to as the 0 direction. The ends of forced with different commercially available polycrys- the tensile bars were encased in a wire mesh to alleviate line SiC-based fibers. The fber types evaluated in this grip stresses and bending moments at and near the udy included (1) the Tyranno ZMI fiber,(2)the Nicalon fiber, (3)the Tyranno SA-3 fiber, and(4) performed along one of the two orthogonal fiber direc the Syl-iBN fiber. In this order, the fiber types typically tions. Room-temperature tensile tests were performed increase in modulus ance. high-tempe using a universal testing machine (Model 8562, capability, and acquisition cost. In addition, MI com- Instron, Canton, MA). Specimens were loaded at a con- posite data reported previously for the Hi-Nicalon stant rate of 4 kN/min. Two clip-on strain gauges(2.5% ype-S fiber, another commercially available high-mod- max strain) were attached, one on each face, and the lus SiC fiber type, are also included in this paper for average of the two strain gauges was used for determin- ing the strain values for the tests. Unload-reload inter ruptions were also performed, usually at least two, in order to determine the residual stress in the composite Experimental Procedure matrIX Modal acoustic emission(AE) was monitored dur- fiber types were produced by 0o ting of four different ing the room-temperature tensile tests. A fracture wave Several fiber preforms consi symmetric lay-up of detector was used with wide- band pass frequency sen- eight plies of 2D-woven five-harness satin fabric with sors(50-2000 kHz), both from Digital Wave Corpo fiber content balanced in the two orthogonal directions. tion(Model B1025, Englewood, CO). Two AE sensors The preforms were then interphase coated with a thin were placed on the face of the specimen, one on each layer of boron nitride by chemical vapor infiltration side of the gauge section, and approximately 50-60 mm CVD), followed by CVI SiC, slurry-cast SiC, and silicon from one another. The two AE sensors were synchro- MI, producing what is commonly referred to 4.5 from the same event at the same time if either sensor was nized, that is, both sensors would record the waveform the slurry-cast melt- infiltration matrix composite Table I lists the panels evaluated in this study, and the triggered. Events that occurred in the gauge section key constituent properties based on in-process panel(25 mm region of the extensometers)were sorted out data and measurements on different specimens from using a threshold voltage crossing technique.and each panel. The panels included (1)three panels rein- used for analysis according to the location of each event forced with Syl-iBN (NASA-treated Syl fiber produced based on the speed of sound of the extensional wave, by Dow Corning, Midland, Mr);(2)three panels with which was determined posttest from events that oc- the Tyranno SA(Ube Industries, Japan);(3)one panel curred between the sensors. .1 Typically, 70% of the with a Hi-Nicalon(Nippon Carbon, Tokyo, Japan); AE activity events occurred outside the gauge section and(4)two panels with the Tyranno ZMI (UBE In- and were not used in the aE analysis
properties such as matrix cracking stresses, ultimate tensile properties, and elevated temperature creep and fatigue properties. Compared with other commercially available fibers, the Syl-iBN fiber evaluated in these studies is very stable against high-temperature degradation both during processing and service, and as a result is expected to be less prone to mechanical performance variation arising from process and/or application variations.6 However, the Syl-iBN fiber is not commercially available readily, and the other fiber types may be more attractive as they offer an overall cost advantage over the Syl-iBN fiber in meeting the necessary property requirements for some applications. The purpose of this study was to assess the in-plane mechanical performance of 2D 0/90 MI composites (oriented in one of the orthogonal fiber directions) reinforced with different commercially available polycrystalline SiC-based fibers. The fiber types evaluated in this study included (1) the Tyranno ZMI fiber, (2) the Hi-Nicalon fiber, (3) the Tyranno SA-3 fiber, and (4) the Syl-iBN fiber. In this order, the fiber types typically increase in modulus, creep resistance, high-temperature capability, and acquisition cost. In addition, MI composite data reported previously5 for the Hi-Nicalon Type-S fiber, another commercially available high-modulus SiC fiber type, are also included in this paper for comparison. Experimental Procedure Several fiber preforms consisting of four different fiber types were produced by 0/90 symmetric lay-up of eight plies of 2D-woven five-harness satin fabric with fiber content balanced in the two orthogonal directions. The preforms were then interphase coated with a thin layer of boron nitride by chemical vapor infiltration (CVI), followed by CVI SiC, slurry-cast SiC, and silicon MI, producing what is commonly referred to as the slurry-cast melt-infiltration matrix composite.1,4,5 Table I lists the panels evaluated in this study, and the key constituent properties based on in-process panel data and measurements on different specimens from each panel. The panels included (1) three panels reinforced with Syl-iBN (NASA-treated Syl fiber produced by Dow Corning, Midland, MI6 ); (2) three panels with the Tyranno SA (Ube Industries, Japan); (3) one panel with a Hi-Nicalon (Nippon Carbon, Tokyo, Japan); and (4) two panels with the Tyranno ZMI (UBE Industries, Tokyo, Japan). In addition, Table I also includes data from two panels with Hi-Nicalon Type-S (Nippon Carbon) fiber that came from the earlier study,5 which is included in this paper for property comparison. For convenience, the composite panels are referred to as xxx-Y where xxx is the reinforcing fiber type (Syl-iBN, SA, HN, ZMI, HNS) and Y is the panel number with that particular fiber. For in-plane mechanical property evaluation, tensile specimens, B150 mm long and 12.6 mm wide at the ends, were machined from the panels into a dog-bone shape where the gauge section length and width were approximately 25 and 10 mm, respectively. The length of each specimen was aligned as close as possible with one of the two orthogonal fiber directions, commonly referred to as the 01 direction. The ends of the tensile bars were encased in a wire mesh to alleviate grip stresses and bending moments at and near the pneumatic pressure grips. All tensile tests were performed along one of the two orthogonal fiber directions. Room-temperature tensile tests were performed using a universal testing machine (Model 8562, Instron, Canton, MA). Specimens were loaded at a constant rate of 4 kN/min. Two clip-on strain gauges (2.5% max strain) were attached, one on each face, and the average of the two strain gauges was used for determining the strain values for the tests. Unload–reload interruptions were also performed, usually at least two, in order to determine the residual stress in the composite matrix.12 Modal acoustic emission (AE) was monitored during the room-temperature tensile tests. A fracture wave detector was used with wide-band pass frequency sensors (50–2000 kHz), both from Digital Wave Corporation (Model B1025, Englewood, CO). Two AE sensors were placed on the face of the specimen, one on each side of the gauge section, and approximately 50–60 mm from one another. The two AE sensors were synchronized, that is, both sensors would record the waveform from the same event at the same time if either sensor was triggered. Events that occurred in the gauge section (25 mm region of the extensometers) were sorted out using a threshold voltage crossing technique7,13 and used for analysis according to the location of each event based on the speed of sound of the extensional wave, which was determined posttest from events that occurred between the sensors.7,13 Typically, 70% of the AE activity events occurred outside the gauge section and were not used in the AE analysis. 152 International Journal of Applied Ceramic Technology—Morscher and Pujar Vol. 6, No. 2, 2009
wwceramics. org/ACT SiC Fiber-Reinforced MI SiC Composites 153 Table I. Composite Phy fber specimen Average f radius [ specimens] Average Average (um)per tow epcm (mm) (scatter) SYLiBN-1 Sylramic-iBN 5 226[1 0.352[11 0.1140.286 (+0.07/-0.19)(+0.014/-0.004) SYLiBN-2 Sylramic-iBN 5 0.386[10] 0.1570.287 (+0.14/-0.12)(+0.026/-0.022) SYLiBN-3 Sylramic-iB 5 800 1.93[10] 0.410[10 0.130.270 0.09 (+0.02/-0.018) SA-1 Tyranno SA3 5 8007.12.057 0.348[7] 0.1200.281 (+0.06/-0.12)(+0.02/-0.01) Tyranno SA3 5 5] 0.3625] 0.281 (+0.04/-0.05)( SA-3 yranno SA3 5 800 215[10 0.332[10 0.098 0.274 (+0.05/-0.08)(+0.006/-0.004) Hi-Nicalon 500 7.1 0.0390.227 (+0.1l/-0.13)(+0.012/-0.01) Tyranno ZMI 5.5 800 3.759 0.227 (+0.004/-0.006 Z-2 Tyranno ZMI 5.5 8.7 362{4] 0.292[4] 0.072 0.198 (+0.12/-0.14)(+0.01/-0.01) HNS-1 Hi-Nicalon S 6.5 5007.12.49团7] 0.3029 (+0.04/-0.09)(+0.012/-0.004) Hi-Nicalon S 6.5 50 217[9 0.3489] 0.040.21 (+0.08/-0.12)(+0.020/-0.018) The preforms were detooled after CVI SiC infiltration. Therefore, the volume of BN could not be measured volume of BN was estimated from average BN thickness measurements of polished specimens. The volume of Sic after CVI infiltration after subtracting the estimated weight of Bn and the known weights of the fibers for the p: i diret d rom lined from the weight gain The fiber. BN, and Cy SiC densities used were 3.05, 1.5, and 3.2 g/cm,, respectively Elevated temperature tensile -rupture tests Results were performed at 1200.C and 1315.C in ambient air on a different machine(Instron Model 5569), which had a resistance-heated MoSiz element furnace inserted into the center of the dog-bone section. The ends of the Table i lists the nominal and calculated values for tensile bars in these tests were also encased in a wire key properties of the constituents in each composite mesh, but the pneumatic grips were water cooled. A panel, based on in-process data and data measured contact extensometer with SiC contacting pins 25 mm on the final processed panels and test specimens apart from one another was used to measure strain at the Because the woven architectures for all panels were bal- edge of the specimen in the gauge section Displacement anced in fiber content in the two orthogonal directions, was measured with an LVDT that featured a maximum the fiber volume fraction in the tensile loading direc- strain capability of 1%. Before the elevated temperature tion, f o, was half of the total fiber volume. For this creep test, a tensile modulus measurement was made on study, fo was determined from the estimated total fiber each specimen over the stress range 5-50 MPa at room area in the loading direction divided by the measured physical area of the composite specimen in the
Elevated temperature tensile creep-rupture tests were performed at 12001C and 13151C in ambient air on a different machine (Instron Model 5569), which had a resistance-heated MoSi2 element furnace inserted into the center of the dog-bone section. The ends of the tensile bars in these tests were also encased in a wire mesh, but the pneumatic grips were water cooled. A contact extensometer with SiC contacting pins 25 mm apart from one another was used to measure strain at the edge of the specimen in the gauge section. Displacement was measured with an LVDT that featured a maximum strain capability of 1%. Before the elevated temperature creep test, a tensile modulus measurement was made on each specimen over the stress range 5–50 MPa at room temperature. Results Constituent Analyses Table I lists the nominal and calculated values for key properties of the constituents in each composite panel, based on in-process data and data measured on the final processed panels and test specimens. Because the woven architectures for all panels were balanced in fiber content in the two orthogonal directions, the fiber volume fraction in the tensile loading direction, fo, was half of the total fiber volume. For this study, fo was determined from the estimated total fiber area in the loading direction divided by the measured physical area of the composite specimen in the gauge Table I. Composite Physical Properties Panel Fiber type Average fiber radius (lm) # of fibers per tow epcm Average specimen thickness (mm) Average f [# specimens] (scatter) Average fBN Average fCVI SiC SYLiBN-1 Sylramic-iBN 5 800 7.9 2.26 [11] 0.352 [11] 0.114 0.286 (10.07/0.19) (10.014/0.004) SYLiBN-2 Sylramic-iBN 5 800 7.9 2.05 [10] 0.386 [10] 0.157 0.287 (10.14/0.12) (10.026/0.022) SYLiBN-3 Sylramic-iBN 5 800 7.9 1.93 [10] 0.410 [10] 0.134 0.270 70.09 (10.02/0.018) SA-1 Tyranno SA3 5 800 7.1 2.05 [7] 0.348 [7] 0.120 0.281 (10.06/0.12) (10.02/0.01) SA-2 Tyranno SA3 5 800 7.1 1.97 [5] 0.362 [5] 0.126 0.281 (10.04/0.05) (70.008) SA-3 Tyranno SA3 5 800 7.1 2.15 [10] 0.332 [10] 0.098 0.274 (10.05/0.08) (10.006/0.004) HN Hi-Nicalon 6.85 500 7.1 3.05 [7] 0.274 [7] 0.039 0.227 (10.11/0.13) (10.012/0.01) Z-1 Tyranno ZMI 5.5 800 8.7 3.75 [9] 0.281 [9] 0.082 0.227 10.06 (10.004/0.006) Z-2 Tyranno ZMI 5.5 800 8.7 3.62 [4] 0.292 [4] 0.072 0.198 (10.12/0.14) (10.01/0.01) HNS-1 Hi-Nicalon S 6.5 500 7.1 2.49 [7] 0.302 [9] 0.04 0.25 (10.04/0.09) (10.012/0.004) HNS-2 Hi-Nicalon S 6.5 500 7.1 2.17 [9] 0.348 [9] 0.04 0.21 (10.08/0.12) (10.020/0.018) The preforms were detooled after CVI SiC infiltration. Therefore, the volume of BN could not be measured directly from weight gain. Instead, the volume of BN was estimated from average BN thickness measurements of polished specimens. The volume of SiC was determined from the weight gain after CVI infiltration after subtracting the estimated weight of BN and the known weights of the fibers for the panel preform. The fiber, BN, and CVI SiC densities used were 3.05, 1.5, and 3.2 g/cm3 , respectively. DiCarlo et al. 6 www.ceramics.org/ACT SiC Fiber-Reinforced MI SiC Composites 153
154 International Journal of Applied Ceramic Technolog-Morscher and pujar Vol.6,No.2,2009 section; that is, is shown for comparison. In Fig. 1, the hysteresis loops f o=(Nply N )(epcm/10)(R)/t(1) were removed for clarity, while Fig. 2 shows represen- tative stress-strain curves with the initial loops and the where Noly is the known number of plies in the lay-up attendant residual stress for the different com- (eight for all the tested in this study); N posite specimens. From these figures and Tables I and the nominal number of fibers per tow: epcm/10 is the II, there are some general fiber-related observations that known tow ends per centimeter of the 2D weave(i.e can be made concerning the as-fabricated number of fiber tows per centimeter) converted to mil specimens. First, as expected from composite theory, limeter; R is the nominal fiber radius in millimeter; and increasing the fiber volume fraction increased the com- t is the measured specimen thickness in millimeter. Ta- posite secondary modulus as well as the ultimate ble l lists the calculated values for the total fiber volume. strength and strain. Second, for the higher modulus fi- bers(Er 380 GPa), increasing the fibe er volume fraction Table I are the nominal N and R values for each iber also increased the composite initial elastic modulus type, as well as the specimen t values and lulus for the Mi matrix in the loading direction is lower than that of the fiber. Third, composite specimen Room-Temperature Stress-Strain Bebavior witb AE with the higher modulus fibers showed that the matrix was under a mild compressive stress(ig. 2 and Table The average room-temperature mechanical proper- ID); in contrast, specimens with approximately the same ties from the stress-strain tests are listed in Table I l, and fraction of the lower modulus fiber showed the matrix some representative stress-strain curves are shown in essentially under zero to a very mild tensile residual stress Fig. 1 for individual specimens from each composite Fourth, for approximately the same fiber fraction, the system. In addition, the stress-strain behavior of an lower modulus fibers exhibited higher composite ultimat HNS-2 composite specimen from Morscher and pujar strain,with the HNS panels being an exception Table Il. Composite Room Temperature Mechanical Properties Average A A UTS (MPa) 8(%) on fibers(GPa) 0.005% AE onset Residual [#RT spec] [ specimens] [ specimens] [#RT spec] offset stress stress stress catter (scatter (s catter (scatter) (MPa) (MPa) (MPa) SYLiBN-1 247 3 0.35{3] 1997[2] 194[3 (+0.007/-0.006)(+36/-32)(+0.04/-0.06(+79-143)(+6/-9) SYLiBN-2 271 2 465[2 0.47[2] 18l[2 189[2]-60[2 (±12) ±0.03 +16 +10 SYLiBN-3 238 1 4[ 0.45[ 176[]155[1]-45[1] SA-1254[ 358[1] 0.33[1] 2000[ 152[ 145[1-20[1 SA-2 236[1] 372[1] 0.34[ 2047[ 178[]138[1 15[ SA-3 230[1 334[l 978[ 178[] 135[1 -30[1] HN 244[7 3l1[ 0.79[7 2272[7] 1266]114团6-46 43/-31)(+17/-10)(+0.12/-0.04)(+208/-141)(+4/-5)(+12/-8)(+7/-8) 213[4] 279[3 0.95[3] 973[4] 111(485[4]+12[4] (+5/-3) 9-6(+0.04/-0.03)(+66/-35)(+7/-6(+10/-15)(+5/-9) 0.83[4] 179[4] 12/-6)(+0.02/-0.03)(+49-53)(+5/-4)(+11/-14)(+8/-7) 1*262[ [ 0.63[1] 2278[1 154[1]150 412[1 147[1 135
section; that is, fo ¼ ðNplyNfÞðepcm=10ÞðpR2 f Þ=t ð1Þ where Nply is the known number of plies in the lay-up (eight for all the composites tested in this study); Nf is the nominal number of fibers per tow; epcm/10 is the known tow ends per centimeter of the 2D weave (i.e., number of fiber tows per centimeter) converted to millimeter; Rf is the nominal fiber radius in millimeter; and t is the measured specimen thickness in millimeter. Table I lists the calculated values for the total fiber volume, f 5 2fo, for all specimens from each panel. Also listed in Table I are the nominal Nf and Rf values for each fiber type, as well as the specimen t values and specimen-tospecimen scatter in these t values. Room-Temperature Stress–Strain Behavior with AE The average room-temperature mechanical properties from the stress–strain tests are listed in Table II, and some representative stress–strain curves are shown in Fig. 1 for individual specimens from each composite system. In addition, the stress–strain behavior of an HNS-2 composite specimen from Morscher and Pujar5 is shown for comparison. In Fig. 1, the hysteresis loops were removed for clarity, while Fig. 2 shows representative stress–strain curves with the initial loops and the attendant residual stress for the different composite specimens. From these figures and Tables I and II, there are some general fiber-related observations that can be made concerning the as-fabricated composite specimens. First, as expected from composite theory,14 increasing the fiber volume fraction increased the composite secondary modulus as well as the ultimate strength and strain. Second, for the higher modulus fi- bers (EfB380 GPa), increasing the fiber volume fraction also increased the composite initial elastic modulus. This is consistent with the hypothesis that the effective modulus for the MI matrix in the loading direction is lower than that of the fiber. Third, composite specimens with the higher modulus fibers showed that the matrix was under a mild compressive stress (Fig. 2 and Table II); in contrast, specimens with approximately the same fraction of the lower modulus fiber showed the matrix essentially under zero to a very mild tensile residual stress. Fourth, for approximately the same fiber fraction, the lower modulus fibers exhibited higher composite ultimate strain, with the HNS panels being an exception. Table II. Composite Room Temperature Mechanical Properties Panel Average E (GPa) [#RT spec] (scatter) Average UTS (MPa) [# specimens] (scatter) Average e (%) [# specimens] (scatter) Average stress on fibers (GPa) [#RT spec] (scatter) 0.005% offset stress (MPa) AE onset stress (MPa) Residual stress (MPa) SYLiBN-1 247 [3] 361 [3] 0.35 [3] 1997 [2] 194 [3] 192 [2] 60 [3] (10.007/0.006) (136/32) (10.04/0.06) (179/143) (16/ 9) 72 77 SYLiBN-2 271 [2] 465 [2] 0.47 [2] 2368 [2] 181 [2] 189 [2] 60 [2] (712) 737 70.03 775 74 716 710 SYLiBN-3 238 [1] 444 [1] 0.45 [1] 2210 [1] 176 [1] 155 [1] 45 [1] SA-1 254 [1] 358 [1] 0.33 [1] 2000 [1] 152 [1] 145 [1] 20 [1] SA-2 236 [1] 372 [1] 0.34 [1] 2047 [1] 178 [1] 138 [1] 15 [1] SA-3 230 [1] 334 [1] 0.30 [1] 1978 [1] 178 [1] 135 [1] 30 [1] HN 244 [7] 311 [7] 0.79 [7] 2272 [7] 126 [6] 114 [6] 4 [6] (143/31) (117/10) (10.12/0.04) (1208/141) (14/5) (112/8) (17/8) Z-1 213 [4] 279 [3] 0.95 [3] 1973 [4] 111 [4] 85 [4] 112 [4] (15/3) (19/ 6) (10.04/0.03) (166/35) (17/6) (110/15) (15/9) Z-2 202 [4] 261 [4] 0.83 [4] 1794 [4] 107 [4] 83 [4] 112 [4] (15/3) (112/6) (10.02/0.03) (149/53) (15/4) (111/14) (18/7) HNS-1 262 [1] 341 [1] 0.63 [1] 2278 [1] 154 [1] 150 20 HNS-2 232 [1] 412 [1] 0.60 [1] 2245 [1] 147 [1] 135 20 DiCarlo et al. 6 154 International Journal of Applied Ceramic Technology—Morscher and Pujar Vol. 6, No. 2, 2009
wwceramics. org/ACT SiC Fiber-Reinforced MI SiC Composites 155 (b)600 00 232GP SYL-iBN fo=020&0.1 400 o=0.18&0.14区 E=210 GPa E=210 GPa E=210 GPa Hysteresis Loops Removed Hysteresis Loops Removed 0 0.20.4 12 Strain. % Fig. I. Representative stress-strain curves from different woven composite systems. Figure 3 shows the ae data from different speci- composite. In essence, the curves in Fig. 3 show the mens for each family of composites, collected during the relative distribution of matrix cracks as stress is increased tensile test. The aE parameter of interest is the energy of in the different composite specimens, and complement E events that occur in the gauge section. A single event the tensile stress-strain data in further understanding was captured on two different sensors. The average en- fiber effects on matrix cracki ergy from each event was determined and used to com- The ae onset stress has been shown to pute the cumulative energy of the events starting from to the onset of fiber-bridged matrix crack formation an the initial event until the final event. Figure 3 shows the is one measure of"matrix cracking stress. "The AE normalized cumulative AE energy(Norm CumAE), onset stress is the onset of a high rate of high-energy AE which is the cumulative energy divided by the total cu- events and determined by extrapolating the steep mulative energy at the final event, plotted versus com- sle tion of the norm Cumae versus stress curve posite stress. It has been shown that for MI composites, back to the zero axis. Table II shows the average values Norm CumAE is directly related to matrix crack den- for the AE onset stress. Also shown are the 0.005% off- sity.The decrease in the rate of Norm CumAE at high set stresses. from the stress-strain curves. a common stress is indicative of matrix crack saturation in the technique for determining the proportional limit, and often associated with matrix cracking strengths for these u0.7 E 02 0.6 Strain. 50100150200250300350400 Fig. 2. Initial part of unload-reload stress-strain curves showing residual stress(circles) for representative specimens from each Fig 3. Acoustic emission behavior during room temperature tensile tests on different fiber-containing MI composites
Figure 3 shows the AE data from different specimens for each family of composites, collected during the tensile test. The AE parameter of interest is the energy of AE events that occur in the gauge section. A single event was captured on two different sensors. The average energy from each event was determined and used to compute the cumulative energy of the events starting from the initial event until the final event. Figure 3 shows the normalized cumulative AE energy (NormCumAE), which is the cumulative energy divided by the total cumulative energy at the final event, plotted versus composite stress. It has been shown that for MI composites, NormCumAE is directly related to matrix crack density.7 The decrease in the rate of NormCumAE at high stress is indicative of matrix crack saturation in the composite. In essence, the curves in Fig. 3 show the relative distribution of matrix cracks as stress is increased in the different composite specimens, and complement the tensile stress–strain data in further understanding fiber effects on matrix cracking. The AE onset stress has been shown to correspond to the onset of fiber-bridged matrix crack formation and is one measure of ‘‘matrix cracking stress.’’7 The AE onset stress is the onset of a high rate of high-energy AE events and is determined by extrapolating the steep slope portion of the NormCumAE versus stress curve back to the zero axis.7 Table II shows the average values for the AE onset stress. Also shown are the 0.005% offset stresses, from the stress–strain curves, a common technique for determining the proportional limit,15 and often associated with matrix cracking strengths for these 0 100 200 300 400 500 600 0 0.2 0.4 0.6 0.8 1 1.2 Strain, % Stress, MPa SA fo = 0.18 & 0.14 [x] SYL-iBN fo = 0.20 & 0.18 ZMI-1 fo = 0.14 E = 210 GPa HN fo = 0.14 E = 220 GPa Hysteresis Loops Removed 0 100 200 300 400 500 (a) 600 (b) 0 0.2 0.4 0.6 0.8 1 Strain, % Stress, MPa SA-2 fo = 0.18 E = 254 GPa SYL-2 fo = 0.20 E = 283 GPa ZMI-1 fo = 0.14 E = 210 GPa HN fo = 0.14 E = 210 GPa Hysteresis Loops Removed HNS-2 fo = 0.17 E = 232 GPa Fig. 1. Representative stress–strain curves from different woven composite systems. –50 0 50 100 150 200 250 300 0 0.2 0.4 0.6 Strain, % Stress, MPa ZMI fo = 0.14 HN fo = 0.14 SA fo = 0.18 fo = 0.2 Syl-iBN Fig. 2. Initial part of unload–reload stress–strain curves showing residual stress (circles) for representative specimens from each composite system. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 50 100 150 200 250 300 350 400 Composite Stress, MPa Norm Cum AE ZMI SA Syl-iBN HN HNS Fig. 3. Acoustic emission behavior during room temperature tensile tests on different fiber-containing MI composites. www.ceramics.org/ACT SiC Fiber-Reinforced MI SiC Composites 155
International Journal of Applied Ceramic Technolog-Morscher and pujar Vol.6,No.2,2009 0005%Ote fractions of fiber, BN, and CVI SiC(Table D). Emini was determined from the rule of mixtures of the fractions of First LOUd Al the three components of the minicomposite where the fiber moduli used were 380 GPa for Syl-iBN, SA, and HNS: 280 GPa for HN; and 200 GPa for ZMI. The bn modulus was 25 GPa, and the CVI SiC modulus was 425 GPa. Plotting NormCumAE versus ominimatris Fig 5 shows essentially that all of the aE behavior of the composites of this study now falls into a much narrower ominimatrix range of <+20 MPa. This reduced range is similar to what was found for the MI composites pro- 140 160180 cessed by a different vendor in a previous study 0. 005% Offset Stress, MPa For predictive modeling, an empirical two-param Fig 4 of0.005% offset stress, AE onset stress and eter Weibull distribution for Norm CumAE was best first Loud AE event for composite specimens tested in this study fitted to the data in Fig. 4 composites. As shown in Fig 4, the AE onset stress and NormCumAE 005% offset stress are close in magnitude; however 1-exp(ominimatrix/ominimatri) the 0.005% offset stress almost always exceeded the AE onset stress for the composites of this study. The AE onset stress is considered to be a better parameter for matrix cracking because it is a direct measure of when where ominimatris is a reference stress and m is the Weibull modulus. For this study, the distribution func- matrix cracks actually occur There is a large scatter in stress over which aE ac tion was only slightly different from that determined in ivity occurs for the different composite systems Morscher6(see Fig. 5). Also note that extrapolating the (+70 MPa for the composites tested in this study- slope of the near-linear region of the distribution rep- Fig 3). It has been shown in Morscher.that the ma- esents a general Ominimatrix for onset of high-energy AE activity, that is, through-thickness matrix crackin trix cracking behavior of 2D woven MI composites with The stress corresponding to the onset of through-thick- similar-sized tows could be correlated to the stress in the region of the composite outside the load-bearing o because these cracks allow relatively easy access for ox- 0 fiber tow and its associated interphase coating and idizing species into the composite and cause environ- CVI SIC matrix. To estimate the in situ stress on the minimatrix? "material, or the portion of the composite This siupgngirine outside of the load-bearing 0" minicomposite, including the 90 minicomposites and the surrounding slurry-cast and MI material, it was shown in Morscher that a sim- a reen= SA H le rule of mixtures relationship could be used 0.6 Purple = ZMI Model from reference 16: E E 1 -mini where oc is the applied composite stress; och is the re- w95 MPa Onset Minirmatrix Siress sidual stress within the as-fabricated composite; Ec is elastic modulus of the composite; Emini is the elastic modulus of the load-bearing 0 f mini is the fraction of 0 minicomposites in the com- Fig. 5. Normalized cumulative acoustic emission energy plotted posite. The composites of this study were all balanced versus the stress in the matrix outside of the load-bearing weaves;therefore, mini is simply half the combined total minicom at 1s. animatrix stress
composites. As shown in Fig. 4, the AE onset stress and 0.005% offset stress are close in magnitude; however, the 0.005% offset stress almost always exceeded the AE onset stress for the composites of this study. The AE onset stress is considered to be a better parameter for matrix cracking because it is a direct measure of when matrix cracks actually occur. There is a large scatter in stress over which AE activity occurs for the different composite systems (770 MPa for the composites tested in this study— Fig. 3). It has been shown in Morscher7,13 that the matrix cracking behavior of 2D woven MI composites with similar-sized tows could be correlated to the stress in the region of the composite outside the load-bearing 01 minicomposite, where the minicomposite consists of the 01 fiber tow and its associated interphase coating and CVI SiC matrix. To estimate the in situ stress on the ‘‘minimatrix’’ material, or the portion of the composite outside of the load-bearing 01 minicomposite, including the 901 minicomposites and the surrounding slurry-cast and MI material, it was shown in Morscher7 that a simple rule of mixtures relationship could be used: sminimatrix ¼ ðsc þ sthÞ Ec Ec fminiEmini 1 fmini ð2Þ where sc is the applied composite stress; sth is the residual stress within the as-fabricated composite; Ec is elastic modulus of the composite; Emini is the elastic modulus of the load-bearing 01 minicomposite; and fmini is the fraction of 01 minicomposites in the composite. The composites of this study were all balanced weaves; therefore, fmini is simply half the combined total fractions of fiber, BN, and CVI SiC (Table I). Emini was determined from the rule of mixtures of the fractions of the three components of the minicomposite where the fiber moduli used were 380 GPa for Syl-iBN, SA, and HNS; 280 GPa for HN; and 200 GPa for ZMI. The BN modulus was 25 GPa, and the CVI SiC modulus was 425 GPa. Plotting NormCumAE versus sminimatrix, Fig. 5 shows essentially that all of the AE behavior of the composites of this study now falls into a much narrower sminimatrix range of o720 MPa. This reduced range is similar to what was found for the MI composites processed by a different vendor in a previous study.7,16 For predictive modeling, an empirical two-parameter Weibull distribution for NormCumAE was best fitted to the data in Fig. 4: NormCumAE ¼ 1 exp ðsminimatrix=so minimatrixÞ m ð3Þ where so minimatrixis a reference stress and m is the Weibull modulus. For this study, the distribution function was only slightly different from that determined in Morscher16 (see Fig. 5). Also note that extrapolating the slope of the near-linear region of the distribution represents a general sminimatrix for onset of high-energy AE activity, that is, through-thickness matrix cracking. The stress corresponding to the onset of through-thickness matrix cracks is an important design parameter, because these cracks allow relatively easy access for oxidizing species into the composite and cause environ- 100 120 140 160 180 200 225 200 175 150 125 100 75 50 0.005% Offset Stress, MPa Stress, MPa 0.005% Offset Stress AE Onset Stress First Loud AE Event Fig. 4. Comparison of 0.005% offset stress, AE onset stress and first Loud AE event for composite specimens tested in this study. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 50 100 150 200 250 300 Minimatrix Stress, MPa Norm Cum AE Blue = Syl-iBN Red = HNS Green = SA Gray = HN Purple = ZMI Fig. 5. Normalized cumulative acoustic emission energy plotted versus the stress in the matrix outside of the load-bearing minicomposites, that is, minimatrix stress. 156 International Journal of Applied Ceramic Technology—Morscher and Pujar Vol. 6, No. 2, 2009
wwceramics. org/ACT SiC Fiber-Reinforced MI SiC Composites 157 mental degradation at elevated temperatures. This onset failure Table ID)multiplied by the volume fraction of ominimatrix, 95 MPa, was about the same as that found fibers in the loading direction(Table D). For the lower in morscher, 16 modulus fiber composites, slightly lower t values were When multiplied by the final matrix crack density, determined as would be expected for smoother surfaces Eq(4)can be used to derive a general relationship for and finer grain sizes of these fiber types. However, for composites. For example, based on Curtin and col- 23+5 MPa. In an earlier study for MI composites fab- strain can be described ricated with lower fiber volume fraction SA lbers the E=o/Ec +adpc/Er(o+oth) best-fit t was found to be between 40 and 50 MPa' For (4) similar Syl-iBN composites from a different vendor, t for pe>28 was found to be x 70 MPa. This difference in t is probably best explained by the thicker BN interphase where the stress-dependent crack density, Pe, can be thicknesses of this study(100 h to
mental degradation at elevated temperatures. This onset sminimatrix, B95 MPa, was about the same as that found in Morscher.16 When multiplied by the final matrix crack density, Eq. (4) can be used to derive a general relationship for matrix cracking in these 2D woven composites and can then be used to model stress/strain behavior for these composites.7 For example, based on Curtin and colleagues,17,18 strain can be described as e ¼ s=Ec þ adrc=Efðs þ sthÞ; for r1 c > 2d ð4Þ where the stress-dependent crack density, rc, can be found from Eq. (3) multiplied by the final crack density and converted back into composite stress, the stressdependent sliding length d ¼ arðs þ sthÞ=2t ð5Þ where r is the fiber radius, t is the interfacial shear stress, and a ¼ ð1 f ÞEm=fEc ð6Þ Em, the elastic modulus of the minimatrix, is assumed to be everything in the composite other than the loadbearing fibers and can be determined from rule of mixtures knowing fo, Ef, and Ec. The only parameter not known is t which can be determined by best fitting Eq. (5) to empirical stress–strain behavior. Figure 1b (circles) shows predicted stress–strain curves using the best-fit t values listed in Table III, which also shows the measured matrix crack densities. The maximum stress circle for the predicted stress– strain curves in Fig. 1b is the average fiber strength at failure (Table II) multiplied by the volume fraction of fibers in the loading direction (Table I). For the lower modulus fiber composites, slightly lower t values were determined as would be expected for smoother surfaces and finer grain sizes of these fiber types. However, for the composites reinforced with the polycrystalline fibers (Syl-iBN and SA), the t values were unexpectedly low at 2375 MPa. In an earlier study for MI composites fabricated with lower fiber volume fraction SA fibers, the best-fit t was found to be between 40 and 50 MPa.5 For similar Syl-iBN composites from a different vendor,7 t was found to be B70 MPa. This difference in t is probably best explained by the thicker BN interphase thicknesses of this study (B1.070.1 mm) compared with the thinner BN interphases used (r0.5 mm) in the other studies.4,5 Using the minimatrix approach described above, it follows that a simple relationship for stress–strain behavior and matrix cracking can be established for a wide range of 2D 0/90 MI woven composite systems. The approach assumes that the matrix porosity is low enough so that matrix cracks emanate from the 901 minicomposites in the 2D architecture and that this cracking mechanism is negligibly dependent on specimen width, length, and thickness (volume effects). It is shown here that this understanding can also be applied to creep-rupture properties, as discussed in the next section. Elevated Temperature Creep-Rupture Behavior Tensile creep-rupture studies were performed in ambient air at 12001C for ZMI, SA, and Syl-iBN composites and at 13151C for SA and Syl-iBN composites. Representative data on total strain versus time data are shown in Fig. 6 at 12001C and 13151C. The creep curves are dominated by a decreasing strain rate. A clear steady state or constant creep-rate region was never achieved in the creep tests used for most specimens in this study. For specimens that failed in rupture during the test, a region of increasing strain rate just before rupture was typically observed. For specimens that survived the creep test, typically B100 h, the test was stopped and either a fast-fracture test was performed at the creep temperature or the specimen was cooled down, removed from the rig, and tested at room temperature, using similar unload–reload and AE monitoring as used for the as-produced specimens. A limited number of specimens were tested for times 4100 h to Table III. Matrix Crack Density and Interfacial Shear Stress Panel Final measured crack density (#/mm) Best fit s (MPa) SYLiBN-1 7.5 25 SYLiBN-2 8.1 18 SYLiBN-3 7.9 20 SA-1 7.8 28 HN 2 9 Z-1 4.1 20 HNS-1 7.0 36 HNS-2 6.4 20 DiCarlo et al. 6 www.ceramics.org/ACT SiC Fiber-Reinforced MI SiC Composites 157
International ournal of Applied Ceramic Techmolog--Morscher and pujar Vol.6,No.2,2009 (a)0.351ZM-1121MPa (b)0.4 0.3//(rupture) 小2172MPa SA-1 172 MF SA-1 138 MP Syl-iBN-3 172 MF SA-1 103 MPa 0.15 詈015 Syl-2 103 MPa SA-1 138 MPa 1200C Creep 315C Creep 100 Time hours Fig. 6. Tensile creep total strain(elastic plus time-dependent)curves at(a)1200C and(b)1315C for diferent fiber-type MI composite The rupture results are plotted as composite stress some of the composite systems in their as-produced versus time at 1200.C and 1315 C in Fig. 7. Note that condition. This stress was calculated using Eq. (3)and a specimens that did not rupture are indicated by arrows. Ominimatrix stress of 95 MPa, as described earlier in Fig In general, the rupture resistance directly correlates to 5. The differences in the composite onset stresses for th the inherent creep and rupture resistance of the fibers, three different fiber composites are a result of differences with Syl-ibN composites being the most, and ZMI in the elastic modulus values for the fiber types as well as composites being the least, creep and rupture resistant. constituent volume fractions, especially fiber volume Also note that there were considerable differences in fi- differences between the composites. At 1200oC, the ber volume fraction between the different composite rupture data(Fig. 7a)show that the predicted onset systems studied, which in effect further accentuated the cracking stress correlates very well with the run-out con differences in the creep performance in these compos- ditions for all three composites systems. Because ites. The Syl- iBN composites, or the composites with the onset of matrix cracks and resultant environmental the most creep-resistant fiber, also had the highest fiber attack through the cracks is the most common mecha volume fractions and ZMI composites, or those with the nism leading to composite rupture under tensile creep, least creep-resistant fiber had the lowest fber volume. this suggests that the room-temperature criterion of (a) a Syl-iBN-3 1315G138M (b)240054N3 1801Hs2 10011315°cce Time. hr Fig.7. Tensile creep behavior of different MI composites at(a)1200C and(b)1315C. Each data point represents an individual specimen tix=95 MPa stress for designated composite p
failure or up to 500 h, followed by retained tensile property measurements at room temperature. The rupture results are plotted as composite stress versus time at 12001C and 13151C in Fig. 7. Note that specimens that did not rupture are indicated by arrows. In general, the rupture resistance directly correlates to the inherent creep and rupture resistance of the fibers, with Syl-iBN composites being the most, and ZMI composites being the least, creep and rupture resistant. Also note that there were considerable differences in fi- ber volume fraction between the different composite systems studied, which in effect further accentuated the differences in the creep performance in these composites. The Syl-iBN composites, or the composites with the most creep-resistant fiber, also had the highest fiber volume fractions and ZMI composites, or those with the least creep-resistant fiber had the lowest fiber volume. Figure 7 also shows the predicted composite stress for the onset of matrix cracking at room temperature for some of the composite systems in their as-produced condition. This stress was calculated using Eq. (3) and a sminimatrix stress of 95 MPa, as described earlier in Fig. 5. The differences in the composite onset stresses for the three different fiber composites are a result of differences in the elastic modulus values for the fiber types as well as constituent volume fractions, especially fiber volume differences between the composites. At 12001C, the rupture data (Fig. 7a) show that the predicted onset cracking stress correlates very well with the run-out conditions for all three composites systems. Because the onset of matrix cracks and resultant environmental attack through the cracks is the most common mechanism leading to composite rupture under tensile creep, this suggests that the room-temperature criterion of 0 0.05 0.1 0.15 0.2 0.25 0.3 (a) 0.35 (b) 0 20 40 60 80 100 120 0 20 40 60 80 100 120 Time, hours Total Strain, % SA-1 138 MPa 1200°C Creep SA-3 155 MPa SA-1 172 MPa (rupture) Syl-iBN -3 209 MPa (rupture) Syl-iBN-3 172 MPa ZMI-1 121 MPa (rupture) ZMI-1 103 MPa 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 Time, hours Total Strain, % 1315°C Creep Syl-2 103 MPa Syl-2 138 MPa Syl-2 172 MPa (rupture) SA-1 138 MPa (rupture) SA-1 103 MPa Fig. 6. Tensile creep total strain (elastic plus time-dependent) curves at (a) 12001C and (b) 13151C for different fiber-type MI composites. 80 100 120 140 160 180 200 220 (a) 240 (b) 0.1 1 10 100 1000 Time, hr Composite Stress, MPa Syl-iBN-3 Syl-iBN-1 SA-3 ZMI-1 ZMI-2 Syl-iBN-3 Pre-crept at 1315C; 138 Mpa Crept to failure at 1200C 1200°C Creep Syl-3 SA-3 ZMI-1 80 100 120 140 160 180 200 220 240 1 10 100 1000 Time, hr Composite Stress, MPa Syl-iBN-3 Syl-iBN-2 Syl-iBN-1 SA-1 SA-2 SA-3 HNS-2 [5] 1315°C Creep Syl-3 SA-3 HNS-2 Fig. 7. Tensile creep behavior of different MI composites at (a) 12001C and (b) 13151C. Each data point represents an individual specimen. Dashed lines represent the sminimatrix 5 95 MPa stress for designated composite panel. 158 International Journal of Applied Ceramic Technology—Morscher and Pujar Vol. 6, No. 2, 2009
wwceramics. org/ACT SiC Fiber-Reinforced MI SiC Composites minimatrix of 95 MPa for matrix crack onset holds for one 100-h interrupted specimen(138 MPa) survived rupture up to 1200C in these composites. At 1315C, 0. 38% total strain. While the 0.3% strain limit is the rupture data(Fig. 7b) show that the predicted onset consistent with the data on both the SA and the Syl-iBN racking stress intersects the rupture curves at x 70 h for fibers, it appears that the ZMi composites in this study I-iBN composites and 10 h for the SA composites. and the hNS composites in the earlier study can with In other words, composite rupture occurs well below the stand higher strains at the lower stresses. In addition, the predicted onset stress derived from the ominimatrix HNS composites in the earlier study featured lower BN 95 MPa criterion. For the HNS composites from and CVI-SiC thickness, which may also have impacted the earlier study, whose data are also included in Fig. these rupture results, and evidently additional data are 7b, the Ominimatrix=95 MPa criterion falls between a needed to better understand the effect of fiber type and 20-h rupture specimen and specimens that survived the underlying mechanisms 100h. Other additional factors such as creep-controlled Based on the data from this study and the earlier matrix crack growth come into play at these tempera- HNS study, two basic design criteria can be formulated cures that contribute to rupture. Nevertheless, the com- for long-time creep properties. The first is based on posite cracking stress limit as predicted by Eq (3)and the stress above which significant matrix cracking the 95 MPa onset minimatrix stress offers a good fir occurs. The composite stress that corresponds to order approach for designing against rupture in air at Ominimatrix=95 MPa gives a design stress limit. Remain- temperatures up to 1200C ing below this stress in a CMC component is essential Figure 8 shows the total strain in the composites for long life. The second is based on the total accumu- plotted versus creep rupture time at 1200.C and lated strain, that is, elastic+time-dependent creep 1315.C obtained from the same creep data at different strain. At applied stress levels below the design stress stress levels. Of particular interest are specimens that limit, rupture life is further limited by the total strain ruptured at stresses below the predicted matrix onset which gives a design strain limit. Maintaining creep con stresses in the creep tests at 1315C. For both the Syl- ditions in the component so that the total strain remains iBN and the SA specimens that were subjected to below this design total strain limit is also essential for 1315.C creep(Fig. 8b)at the lower stresses, the strain long life. The value of this design total strain limit is to failure for rupture was approximately 0.3%. This dependent on the fiber type in the composite. For both may imply a creep strain limit for the initially uncracked the Syl-iBN and the SA composites, this total strain or microcracked(no fiber-bridged cracks) matrix con- limit is about 0. 3% total (elastic+ time-dependent dition. Very little data are available for the HNS fber at strain. While additional testing and modeling efforts 1315%C, and for all fiber types in the tests at 1200%C. are required to understand the strain accumulation for For the HNS composite tested at 1315.C(Fig. 8b), the variable stress states and long times, the results from this sole rupture specimen failed at -0.55% total strain and study can be used to derive some simple models for (b)06 0.45 ●HNs 0.15 1200C Creep Time. hr Fig 8. Tensile creep behavior, total strain versus time, of different MI composites at(a)1200 Cand (b)1315C. Each data point represents
sminimatrix of 95 MPa for matrix crack onset holds for rupture up to 12001C in these composites. At 13151C, the rupture data (Fig. 7b) show that the predicted onset cracking stress intersects the rupture curves at B70 h for Syl-iBN composites and B10 h for the SA composites. In other words, composite rupture occurs well below the predicted onset stress derived from the sminimatrix 5 95 MPa criterion. For the HNS composites from the earlier study, whose data are also included in Fig. 7b, the sminimatrix 5 95 MPa criterion falls between a 20-h rupture specimen and specimens that survived 100 h. Other additional factors such as creep-controlled matrix crack growth come into play at these temperatures that contribute to rupture. Nevertheless, the composite cracking stress limit as predicted by Eq. (3) and the 95 MPa onset minimatrix stress offers a good firstorder approach for designing against rupture in air at temperatures up to 12001C. Figure 8 shows the total strain in the composites plotted versus creep rupture time at 12001C and 13151C obtained from the same creep data at different stress levels. Of particular interest are specimens that ruptured at stresses below the predicted matrix onset stresses in the creep tests at 13151C. For both the SyliBN and the SA specimens that were subjected to 13151C creep (Fig. 8b) at the lower stresses, the strain to failure for rupture was approximately 0.3%. This may imply a creep strain limit for the initially uncracked or microcracked (no fiber-bridged cracks) matrix condition. Very little data are available for the HNS fiber at 13151C, and for all fiber types in the tests at 12001C. For the HNS composite tested at 13151C (Fig. 8b), the sole rupture specimen failed at B0.55% total strain and one 100-h interrupted specimen (138 MPa) survived 0.38% total strain. While the B0.3% strain limit is consistent with the data on both the SA and the Syl-iBN fibers, it appears that the ZMI composites in this study and the HNS composites in the earlier study can withstand higher strains at the lower stresses. In addition, the HNS composites in the earlier study featured lower BN and CVI-SiC thickness, which may also have impacted these rupture results, and evidently additional data are needed to better understand the effect of fiber type and the underlying mechanisms. Based on the data from this study and the earlier HNS study, two basic design criteria can be formulated for long-time creep properties. The first is based on the stress above which significant matrix cracking occurs. The composite stress that corresponds to sminimatrix 5 95 MPa gives a design stress limit. Remaining below this stress in a CMC component is essential for long life. The second is based on the total accumulated strain, that is, elastic1time-dependent creep strain. At applied stress levels below the design stress limit, rupture life is further limited by the total strain, which gives a design strain limit. Maintaining creep conditions in the component so that the total strain remains below this design total strain limit is also essential for long life. The value of this design total strain limit is dependent on the fiber type in the composite. For both the Syl-iBN and the SA composites, this total strain limit is about 0.3% total (elastic1time-dependent) strain. While additional testing and modeling efforts are required to understand the strain accumulation for variable stress states and long times, the results from this study can be used to derive some simple models for 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 (a) 0.5 (b) Time, hr Total Strain, % Syl-iBN-3 Syl-iBN-1 SA-3 ZMI-1 ZMI-2 1200°C Creep 0 0.1 0.2 0.3 0.4 0.5 0.6 0.1 1 10 100 1000 1 10 100 1000 Time, hr Total Strain, % Syl-iBN-3 Syl-iBN-2 Syl-iBN-1 SA-1 SA-2 SA-3 HNS 1315°C Creep 0.31 % total strain for Syl-iBN and SA Fig. 8. Tensile creep behavior, total strain versus time, of different MI composites at (a) 12001C and (b) 13151C. Each data point represents an individual specimen. www.ceramics.org/ACT SiC Fiber-Reinforced MI SiC Composites 159
International Journal of Applied Ceramic Technolog--Morscher and Pujar Vol 6, No 2, 2009 eep. Much more effort is needed in this area to model the elastic modulus at temperature was on average 10% he cause of transient creep and the effect of stress-time less than that measured at room temperature history, as well as to understand the effect of constituent Figure 9 shows that one consequence of creep at content variation on these properties stresses below which fiber-bridged matrix cracking curs was that specimens exhibited a higher stress for Residual Properties after Creep nonlinearity and through-thickness cracking after creep compared with as-produced composites, both at room S temperature and at elevated temperature. This was ob- test were tested to determine their residual stress-strain served for all three fiber-type composites in this study behavior. Specimens were unloaded from the creep test and as reported previously also for the HNS compos- at temperature and were either immediately reloaded to ites. The most dramatic increases were for the most failure at the creep temperature or cooled(with no load) creep-resistant fiber-type, Syl-iBN, composites. As re- and tested at room temperature where aE was moni- ported previously, the cause of this increase is due to red during the test, and two or three unload-reload stress relaxation in the matrix, particularly the Si-SiC loops were performed at increasing stress levels until particulate portion. This results in an increased com- composite failure. Some examples of the stress-strain pressIve stress in the matrix upon unloading, whic behavior after creep are shown in Fig 9 along with m ust be overcome to form bridged -matrix cracks 19,20 room-temperature and high-temperature fast fracture As a further validation to this hypothesis, a Syl-iBN-3 curves. The tensile curves for the after-creep specimens composite specimen was precrept for 50 h at 1315"C are offset by the permanent deformation acquired dur- and 138 MPa. This was then crept at 1200C at ing creep in the stress-strain plots in Fig 9. These plots 220 MPa(as noted in Fig. 7a). The precept specimen therefore show the total accumulated mechanical strain ruptured after 58 h of creep compared with only 0.3 h before failure for the various as-produced and crept for the as-produced, no precreep, specimen, which rep specimens. In all cases of this study, the total accumu- resents an improvement of nearly two orders of magni- lated strain-to-failure is always less than the room-tem- tude. With respect to stress, the precreep condition perature strain-to-failure in the as-produced specimen enabled this to withstand a 25 MPa higher Little change was observed in the elastic modulus for the retained strength tests whether tested at room temper stress than what would be expected for a virgin st for men,a 13% improvement. This of course was only ature or at elevated temperature. It should be noted that oahe specimen; however, it is consistent with all the other 1315C Fast Fracture posed underlying mechanisms. This test also demon- strates that this concept offers the potential for boosting rupture life or rupture stress along the primary fiber di rections for short time(<100 h)applications. The residual ultimate tensile strength(UTS)of the 1315C Creep followed by RT Fast Fracture crept composites is plotted versus total strain for spec- Imens crept at1200°Cand1315° C in Fig.10ina normalized form. This was done because composites hital la dine 2/42: Eater creep i=243 Pa varied in fiber volume fraction and composites of the same fiber type may vary in as-produced strengths due to processing or fiber-lot variation(Table In). There is the data and overlap between the ep for Syl-iBN-2 composit tained strengt Some tests erformed at the creep temperature immediately ues at the creep temperature com- after the creep test(unload, then reload to failure). Other tests were pared with those at room temperature. However, there performed at room tempe is relatively good agreement between the different com- hysteresis loops(not shown). Also included are room temperature posites. In general, under fast-fracture conditions, the unload-reload stress strain and elevated stress-strain curves as well 1200oC and 1315C UTS values are 75% of the as the creep curves in the strain domain. Note that the hysteresis room temperature UTS for both the as-produced and loops have been removed. crept osites(Figs. 10a and b), and the 75%
creep. Much more effort is needed in this area to model the cause of transient creep and the effect of stress-time history, as well as to understand the effect of constituent content variation on these properties. Residual Properties after Creep Specimens that did not rupture during the creep test were tested to determine their residual stress–strain behavior. Specimens were unloaded from the creep test at temperature and were either immediately reloaded to failure at the creep temperature or cooled (with no load) and tested at room temperature where AE was monitored during the test, and two or three unload–reload loops were performed at increasing stress levels until composite failure. Some examples of the stress–strain behavior after creep are shown in Fig. 9 along with room-temperature and high-temperature fast fracture curves. The tensile curves for the after-creep specimens are offset by the permanent deformation acquired during creep in the stress–strain plots in Fig. 9. These plots therefore show the total accumulated mechanical strain before failure for the various as-produced and crept specimens. In all cases of this study, the total accumulated strain-to-failure is always less than the room-temperature strain-to-failure in the as-produced specimen. Little change was observed in the elastic modulus for the retained strength tests whether tested at room temperature or at elevated temperature. It should be noted that the elastic modulus at temperature was on average 10% less than that measured at room temperature. Figure 9 shows that one consequence of creep at stresses below which fiber-bridged matrix cracking occurs was that specimens exhibited a higher stress for nonlinearity and through-thickness cracking after creep compared with as-produced composites, both at room temperature and at elevated temperature. This was observed for all three fiber-type composites in this study, and as reported previously5 also for the HNS composites. The most dramatic increases were for the most creep-resistant fiber-type, Syl-iBN, composites. As reported previously,5 the cause of this increase is due to stress relaxation in the matrix, particularly the Si–SiC particulate portion. This results in an increased compressive stress in the matrix upon unloading, which must be overcome to form bridged-matrix cracks.19,20 As a further validation to this hypothesis, a Syl-iBN-3 composite specimen was precrept for 50 h at 13151C and 138 MPa. This was then crept at 12001C at 220 MPa (as noted in Fig. 7a). The precrept specimen ruptured after 58 h of creep compared with only 0.3 h for the as-produced, no precreep, specimen, which represents an improvement of nearly two orders of magnitude. With respect to stress, the precreep condition enabled this specimen to withstand a 25 MPa higher stress than what would be expected for a virgin specimen, a 13% improvement. This of course was only for one specimen; however, it is consistent with all the other observations on the after-creep properties and the proposed underlying mechanisms. This test also demonstrates that this concept offers the potential for boosting rupture life or rupture stress along the primary fiber directions for short time (o100 h) applications. The residual ultimate tensile strength (UTS) of the crept composites is plotted versus total strain for specimens crept at 12001C and 13151C in Fig. 10 in a normalized form. This was done because composites varied in fiber volume fraction and composites of the same fiber type may vary in as-produced strengths due to processing or fiber-lot variation (Table II). There is some scatter in the data and overlap between the retained strength values at the creep temperature compared with those at room temperature. However, there is relatively good agreement between the different composites. In general, under fast-fracture conditions, the 12001C and 13151C UTS values are B75% of the room temperature UTS for both the as-produced and the crept composites (Figs. 10a and b), and the 75% 0 100 200 300 400 500 0 0.1 0.2 0.3 0.4 0.5 Strain (%) Stress (MPa) Syl-iBN-2 Fig. 9. Residual properties after creep for Syl-iBN-2 composites. Some tests were performed at the creep temperature immediately after the creep test (unload, then reload to failure). Other tests were performed at room temperature with several unload–reload hysteresis loops (not shown). Also included are room temperature unload–reload stress strain and elevated stress–strain curves as well as the creep curves in the strain domain. Note that the hysteresis loops have been removed. 160 International Journal of Applied Ceramic Technology—Morscher and Pujar Vol. 6, No. 2, 2009