311?... 93...; I26... v. at... ti .1 D shift? I. .33.... . Eat. .ZLEnLu. . . , ... .\ V” 1...,‘..5__m._§mé;. . _ _ A . ‘ 2.3:. .27. V-fiai: :3. fit) lHllIHIIHHIIIIllIIHHIHIililllillillllllhUlillllllllhl 3 1293 01037 3664 This is to certify that the dissertation entitled Optimization of Metal-Ceramic Bonding to Enhance Mechanical Properties of Ceramic Matrix Composite presented by Taekoo Lee has been accepted towards fulfillment of the requirements for Ph.D. degree in Materials Science 2 Major professor Date 9-29-93 MS U i: an Affirmative Action/Equal Opportunity Institution 0- 12771 LIBRARY Michigan State University PLACE IN 38111th BOX to remove this checkout from your record. TO AVOID FINES Mum on at bdoro dd. duo. DATE DUE DATE DUE DATE DUE Optimization of Metal-Ceramic Bonding to Enhance Mechanical Properties of Ceramic Matrix Composite by Taekoo Lee A DISSERTATION submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Materials Science and Mechanics 1993 K. N. Subramanian, Advisor Abstract Optimization of Metal-Ceramic Bonding to Enhance Mechanical Properties of Ceramic Matrix Composite by Taekoo Lee The role of interface on the fracture toughness of ceramic glass reinforced with metallic ribbon was investigated. A strong interfacial bonding strength is essential for the load transfer between the matrix and ribbon, and for the utilization of the ductility of the reinforcement. However, weak interfacial bonding is considered to be important so that ribbon pull-out can improve the toughness of brittle ceramic matrix composites. The main aim of the present study is to carry out interface design to optimize mechanical properties of brittle ceramic matrix composites reinforced with ductile metallic ribbons. Acknowledgment The author wishes to express his sincere gratitude to his major thesis advi- sor Dr. K.N. Subramanian for his generous assistance and guidance which ultimately led to the completion of the work reported in this dissertation. The author is also grateful to the members of his doctoral committee, Dr. David S. Grumman, Dr. Dashin Liu and Dr. Carl Foiles, who with their helpful and understanding attitude greatly encouraged him in the course of his research. Thanks are also extended to the Composite Materials and Structures Cen- ter at Michigan State University for financial support during the doctoral program. Finally, and above all, I wish to express my deepest thanks to my parents, my lovely wife, Suekyung, and son, Kwangmin, for their consistent sup- port, encouragement and understanding, throughout the course of this study. ii Contents LIST OF TABLES LIST OF FIGURES 1. Introduction 2. Literature Survey 2-1. Evaluation of Interfacial Bonding Strength in Ceramic Matrix Composite 2-1-1. Interface Test Methods 2-1-2. Theoretical Analyses of Interfacial Debonding and Fiber Pull-out during Interface Tests ------------ 2-2. Fracture Toughness Measurements in Ceramic Matrix Composites 2-3. Strength Analyses in Ceramic Matrix Composites ----- 3. Experimental Procedure 3-1. Materials 3-2. Composite Fabrication 3-3. Interface Modification 3-3-1. Thermal Cycling Technique 3-3-2. Acid Etching Technique 3-4. Mechanical Testing 3-5. Micrographic Observation 4. Results and Discussion 4-1. Theoretical and Experimental Analysis of Debonding Pro- cess of Ductile Ribbon Embedded in Brittle Matrix 4-1-1. Theoretical Modelling iii H 11 17 21 24 25 29 32 32 36 37 4-1-1-1. Load Required to Initiate Debonding ------- 4-1-1-2. Load Required for Continuation of Debond- ing 4-1-2. Experimental Analysis 4-2. Interface Modifications 4-2-1. Thermal Cycling Technique 4-2-1—1. Effects of Thermal Cycling on The Interfa- cial Bonding Strength and Frictional Shear Stress 4-2-1-2. Effects of Thermal Cycling on The Com- plete Debonding Stress 4-2-2. Acid Etching of The Ribbon 4-3. Fracture 'lbughness Behavior 4-3-1. Effect of Fracture Toughness(K1c) on Interfacial Bonding 4-3-2. Tensile Behavior of Pre-cracked Composite ----- 4-3-3. Toughness due to Crack Bridging(AGc) ------- 5. Conclusions 6. References iv 38 40 44 56 67 73 87 92 94 100 103 Table 1. Table 2. Table 3. Table 4. Table 5. Table 6. Table 7. List of Tables Physical and mechanical properties of constituents of the composite. Measured interfacial properties of 650-TC specimens after various numbers of thermal cycles. ----------- Measured interfacial properties of thermal cycled speci- mens up to various maximum temperatures after 5 ther- mal cycles. Fracture toughness (ch) of soda-lime glass matrix rein- forced Nichrome ribbon with various bonding conditions. Effect of interfacial bonding on the displacement at the ribbon fracture. Fracture toughness due to crack bridging (AGc) in vari- ous bonding conditions. Critical embedded length (19) for the maximum fracture toughness due to crack bridging in various bonding con- ditions. 27 62 63 88 93 95 98 Fig. Fig. Fig. Fig. 8 List of Figures A schematic illustrating mechanisms of toughening in ceramic matrix reinforced with ceramic fibers. ------- A schematic illustrating mechanisms of toughening in ceramic matrix reinforced with metallic fibers. -------- Schematic diagrams of indentation techniques: (a) with microhardness indenter(thick specimen), ----- (b) with spherical indenter(thick specimen) ---------- (c) push-through specimen (thin slice of the composite). Schematic diagram of the fiber push-through and push- back apparatus. (a) Variation in average interfacial shear stress with embedded fiber length. (b) Variation in maximum pull-out load with embedded fiber length. Variation of maximum pull-out load: (a) as a function of embedded fiber length, -------- (b) as a function of fiber displacement, for various friction conditions. Dugdale-Barenblatt crack model: (a) Cohesive zone at crack tip with restraining stress dependent on separation distance. (b) Force-displacement relation for atomic attraction in elastic brittle fracture. Typical stress-strain curve of the Nichrome ribbon 10 13 13 14 14 2O 20 Fig. 9 Fig. Fig. .10 11 .12 .14 .16 .17 18 obtained from tensile test. The schematics of sandwiching of a series of aligned rib- bons between the glass slides: a) general set-up b) composite from which pull-out test samples were cut out c) composite from which fracture toughness test samples were cut out The schematic Of typical pull-out specimen. ............ The schematics of typical fracture toughness test speci- mens: a) single notched specimen for flexure test .............. b) pre-cracked specimens for tensile test. Schematics of experimental details of the tensile test: a) general set-up b) self-aligning fixture. Shear stresses due to tensile loading at edge of the embedded area. Load (P) required for continuation of debonding. -------- Static frictional force at the debonded interface. ---------- Typical load-displacement curve obtained in pull-out tests of Nichrome ribbon embedded in soda lime glass matrix. SEM micrograph of the partially debonded interface between the Nichrome ribbon and the soda lime glass matrix. Mechanical keying of debonded surfaces produced by vii 26 28 28 28 30 31 31 33 33 39 41 43 45 47 Fig. Fig. Fig. Fig. Fig. 19 20 21 22 .23 .24 .25 .26 27 roughness of debonded surfaces, transverse contraction, and longitudinal plastic extension of the ribbon due to tensile loading after debonding. SEM micrograph of the interface between the Nichrome ribbon and soda lime glass matrix after debonding progresses to some degree. Interfacial shear stresses for debonding various embed- ded lengths of the Nichrome ribbon. Loads required to continue debonding as a function of debonded length. Effect of embedded length on the normal stress on the ribbon when debonding starts (6,), and normal stress when debonding is completed (Cf) in soda lime slide glass reinforced with Nichrome ribbon. Typical stress-displacement curves of as-fabricated, and 650-TC specimens after various numbers of thermal cycles. SEM micrographs at interfaces of: (a) as-fabricated specimens, (b) 650-TC specimen after 3 thermal cycles, (c) 650-TC specimen after 5 thermal cycles. Stress-displacement curves of specimens that have undergone 5 thermal cycles with various maximum tem- peratures. Interfacial bonding strengths of 650-TC specimens after various numbers of thermal cycles. The effect of maximum cyclic temperature on the interfa- 49 50 51 52 55 57 58 59 60 61 65 Fig. 28 Fig. 29 Fig. 31 Ifig. 32 cial bonding strength after 5 thermal cycles. ------------ The complete debonding stresses as a function of the embedded ribbon length after various numbers of ther- mal cycles of 650-TC specimens. SEM micrographs at debonded interface: (a) due to thermal stress, (b) due to shear stress during pull-out tests. --------- The effect of maximum cyclic temperature on the com- plete debonding stress. Ribbon surfaces roughened by acid etching of: a) as-received, b) pre-oxidized ribbons for 90 seconds The effect of ribbon surface treatment on stress-displace- _ ment curves obtained from pull-out tests. ............. Fig. 33 Fig. 34 Fig. 35 Debonded surface in which debonding crack has propa- gated along the interface between metal and oxide layer. SEM micrographs of the ribbon and the matrix surfaces after ribbon is completely pulled-out from the matrix for the composite with: a) the ribbon roughened by acid etching for 60 seconds, (ribbon) (martix) b) the pre-oxidized ribbon roughened by acid etching for 60 seconds. (ribbon) (matrix) Schematic profile of the interface after ribbon surface treatment, composite fabrication, and during pull-out 66 68 69 7O 72 74 75 76 77 79 8O 81 82 Fig. 36 Fig. 37 Fig. 38 Fig. 39 test for the composite with: a) as-received ribbon roughened by etching, -------- b) pre-oxidized ribbon roughened by etching. -------- Pull-out stress of ribbons that have undergone various treatments for an embedded length of 1.7 mm. ------- Fractographic observation of the crack shielding by the ribbon. Interference fringes represent the path of the crack. Typical stress-displacement curves obtained from the tensile loading of the pre-cracked composite specimens with various interfacial bonding conditions. -------- Stress-displacement curves showing the efl'ect of the embedded ribbon length on toughness (AGO) of NT com- posite specimens. 83 84 86 9O 91 97 1. Introduction Ceramics are a promising class of structural materials for applications where high strength, high stiffness, low thermal expansion, low density and high temperature stability are desirable attributes. However, the brit- tleness of ceramics which catastrophically fail is always a problem in use of conventional ceramics as a structural materials. Ceramic matrix compos- ites may achieve the marked improvements in mechanical properties, espe- cially in fracture toughness, by reinforcements. The main goal of ceramic matrix composites is to reduce the brittleness of ceramics while retaining most of the strength of the ceramic matrix, for structural applications. Various studies on fracture toughness and interfacial bonding in ceramic matrix reinforced with ceramic fibers have shown that high frac- ture toughness can be obtained in ceramic matrix composites that possess weak interfacial bonding (1-15). A major mechanism of enhancement of fracture toughness in such a composite system is the bridging of crack sur- faces by intact fibers (Fig. 1). This involves the frictional sliding of fibers against the matrix following interfacial debonding and failure of the fibers. It requires weak interfacial bonding between the reinforcements and matrix. On the other hand, ceramic matrix composites containing metallic reinforcements strongly bonded to the matrix have exhibited profoundly high fracture toughness. The work done during plastic deformation of the metallic reinforcements between the crack surfaces in the bridging zone contributes to the overall toughness of these composites (16-28) (Fig. 2). Therefore, if both plastic deformation of the reinforcement and frictional sliding between the components can be utilized in the same composite, the MATRIX crack front Fig.1 A schematic illustrating mechanisms of toughening in ceramic matrix reinforced with ceramic fibers. frictional sliding plas c deform ti“- 'd bondi g' Fig. 2 A schematic illustrating mechanisms of toughening in ceramic matrix reinforced with metallic fibers. fracture toughness of the ceramic matrix composite with metallic reinforce- ment may be significantly improved. Since such contributions are con- trolled by the interfacial bonding strength between the matrix and the reinforcement, the maximum fracture toughness can be achieved by modi- fication of interfacial characteristics. Several investigations of interfacial shear stress on fiber pull-out in ceramic matrix composites have been reported (29-36). However, the plas- tic behavior of the metallic fibers in brittle ceramic matrix composites has not been taken into account in analyses of fiber pull-out reported in litera- ture. Slip due to plastic defamation of the metallic fiber may affect the distribution of shear stress at the bonded interfacial area. Moreover, the radial plastic contraction of the metallic fiber at tensile stresses beyond yield stress may take its surface out of contact with the matrix resulting in significantly reduced residual gripping pressure. Coating of the reinforcements is the most widely used technique to mod ify the interfacial characteristics in ceramic matrix composites (37-40). Modification of the interface between the reinforcement and matrix by means of thermal cycling is also a possibility when the thermal expansion mismatch between the components is relatively large. Radial and tangen- tial stresses are generated at the interface of the composite during cooling after composite fabrication due to the mismatch of thermal expansion coef- ficients between the reinforcement and matrix. These residual thermal stresses may cause microcracking at the interface or in the matrix, depend- ing on the interfacial bonding strength and matrix strength of the compos- ite. In glass matrix composites, microcracks present in the matrix can heal by viscous flow during heating portion of thermal cycling (41). However, microcracks in the interface layer can not be expected to self-heal. Instead, with continued thermal cycling, such interfacial microcracks will link together, resulting in weaker interfacial bonding. Interface modification can also be achieved by surface treatments such as chemical etching of the metallic reinforcements in as-received or pre- oxidized condition prior to the composite fabrication. Moore and Kunz (42) showed that in glass matrix composites containing metal particles, acid etching of pre-oxidized metal particles generates extensive, uniform sur- face roughness and that the cracks proceed along the roughened contour of the particles during the composite fracture. As a result, fracture tough- ness in this composite considerably improved with such a surface treat- ment of metal particles. The principal aim of present study is to elucidate interfacial effects on mechanical properties of brittle matrix with ductile reinforcement, and finally to control the interfacial bonding strength in order to obtain opti- mum mechanical properties in the composite. Ribbon geometry of the rein- forcement was selected in this study to facilitate ease of fabrication and theoretical modelling of the composite. First of all, a method to evaluate the interfacial bonding strength of this composite, single ribbon pull-out test was adopted in the present study to investigate the interfacial effects in this composite. Secondly, the interfacial characteristics between the rib- bon and matrix were modified to obtain the optimum interfacial bonding in this composite by means of thermal cycling and acid etching techniques. Finally using the specimens with various interfacial bonding conditions achieved, increment in fracture toughness of the composite due to crack bridging was evaluated and consequently, optimum interfacial bonding strength for maximum fracture toughness in this composite was deter- mined. 2. Literature Survey 2-1. Evaluation of Interfacial bonding strength in Ceramic Matrix Composite 2-1-1. Interface Test Methods Due to importance of the interfacial properties in composites, numerous studies have focussed on controlling interfacial bonding strength by means of modifying interfacial characteristics to achieve desirable composite properties. Therefore, there is considerable interest in developing simple experimental techniques and adequate model for assessing interfacial 1 properties. It may be possible to observe debonding initiation in trans- verse tension and calculate the bonding strength from micromechanics in some metal matrix composites (43). However, for most composites, this technique is not adequate because composite fracture precedes interfacial debonding (44). The single fiber pull-out test has been widely used to assess interfacial properties of composites (29-36, 45-53). The pull-out test can determine the interfacial bonding strength from not only the stress conditions when debonding initiates, but also the frictional sliding resistance as the deb- onded fiber is pulled out from the matrix. However, this technique has dif- ficulty in fabrication of the pull-out test specimens in ceramic matrix composites, and also the interface in specially prepared pull-out specimens may not be representative of the actual composites. To avoid the difficulties associated with the pull-out tests in ceramic matrix composite, Marshall has used microindentation method [Fig. 3(a)], which is referred to as push-out (or push-down) tests (54,55), to measure interfacial properties of glass ceramic composites reinforced with SiC fibers. This technique involves application of a force to the end of a fiber embedded in a matrix using a microhardness indenter. Interfacial shear stresses can be evaluated from the applied load and the displacement of the fiber. Grande et al. (44) have used a spherical indenter [Fig 3(b)] instead of conventional Vickers microhardness indenter in order to get rid of the problem arising from sharp indenter; at high load, cracks may be introduced into the fibers that can influence the measurements. Since such indentation techniques use real composites, the problems involved in single fiber pull-out tests do not arise in these techniques. However, the identification of the debonding event involves a somewhat inaccurate pro- cedure. Laughner et al. (56) have used thin slices of the composite speci- mens instead of thick specimens as a variation of the above techniques, which is sometimes referred to as a push-through method [Fig. 3(c)]. The thin slice is advantageous for frictional measurements for known debonded length; however, sufficiently thin slices are difficult to prepare properly and test, for small-diameter fiber systems. Recently an improved fiber push-through test has been designed by Warren et al. (57) in metal and glass matrix composites reinforced with SiC fibers (Fig. 4). This technique involves push-back procedure following push-through test, which can give additional information about frictional sliding characteristics of interface in such composites. Since indentation techniques are not suitable for this study due to the geometry of the ribbon reinforcement, the single ribbon pull-out test was used in this study to evaluate interfacial behavior. Load lndenter (8) Matrix Fiber S herical 0’) lanenter Fig. 3 Schematic diagrams of indentation techniques (a) with microhardness indenter (thick specimen), (b) with spherical indenter (thick specimen), and (c) push-through specimen (thin slice of the composite). i- -> <- Hf I 2. L i '2 Fiber //A (C) Fig. 3 (continued) Matrix A Base 10 Cylindrical Push Rod Fiber Matrix Piston Support Base Fig. 4 Schematic diagram of the fiber push-through and push-back apparatus. 11 2-1-2. Theoretical Analyses of Interfacial Debonding and Fiber Pull-out during Interface Tests A number of investigations of interfacial shear stress on fiber pull-out in composites have reported so as to evaluate the bonding strength between fiber and matrix, using above interface test methods (29-36). Kelly and Tyson (45) have assumed that shear stress at the interface is distributed uniformly along the fiber embedded into metal matrix. The stress in the fiber required to pull-out the fiber from metal matrix is given by of = 218; (Eq. 1) where I, is the shear stress at the fiber/matrix interface, a: is a distance from the end of the fiber, and r is the radius of the fiber. Since this assumption is based on the idea of metal-matrix yielding under high shear stress, it can not be applied to ceramic matrix composite in which the shear stresses at the interface are not alleviated by matrix yielding. A general relationship between fiber/matrix interfacial shear stress and embedded fiber length in a pull-out specimen has been developed by Greszczuk (29), using the assumptions of shear-lag theory (58). The equi- librium of forces acting on an element of the fiber in a model pull-out spec- imen and the corresponding boundary conditions are considered. As a result, he has derived the maximum shear stress along the interface occur- ring at the point where the fiber enters the matrix, which is given by Tmax = ale cothale (Eq. 2) av 12 where can is the average shear stress along the interface, I, is embedded length of the fiber, and a is a shear-lag parameter dependent on elastic properties of the constituents and their dimensions (58). If the maximum shear stress exceeds the interfacial shear strength of the composite, imme- diate catastrophic debonding between the fiber and the matrix occurs. Lawrence (30) has developed a theory taking partial debonding and the effect of frictional forces acting over the debonded portion of the interface into account. The interfacial frictional force along the debonded interface results from the existence of residual gripping pressure. The applied load should overcome this static frictional force along the debonded interface in addition to the force required for debonding the bonded regions of the interface. Using the same geometry of the model and the results as Greszczuk’s, he discussed the maximum load (Pflax) on the fiber required to achieve complete debonding and initiate pull-out from the matrix. He concluded that whether debonding continues at this load or an increase in load is necessary depends on the embedded length of the fiber (1,) and the ratio of the interfacial shear strength (1;) to the frictional shear stress after debonding (17). In comparison with Greszczuk’s results (Fig. 5), his conclu- sion is illustrated in Fig. 6. Lawrence assumed that the interfacial frictional shear stress(1:f) is con- stant over the debonded region. However, Takaku and Arridge (31) have pointed out that due to the Poisson’s effect, the fiber contracts in the lat- eral direction under tensile stress. This contraction reduces the compres- sive stress from the matrix, and consequently the interfacial frictional force decreases. Although they have noticed such behavior of interfacial frictional stress, they, like Greszczuk, did not take the contribution of fric- tional shear stress along the debonded interface to their maximum load 13 1iav Tmax p 'e (a) meax‘ - n n // rav— co sta t ‘\ two.) > '9 (b) Fig. 5 (a) Variation in average interfacial shear stress with embed- ded fiber length. (b) Variation in maximum pull-out load with embedded fiber length [from Greszczuk (29)]. 14 pma f ’X ,r'lv'i/Tf=1 / /l1 ocle (a) Pf‘ Ti/‘Cf=1 Pf‘ 1 p > 6 5 8 (b) Fig. 6 Variation. of maximum pull-out load (a) as a function of embedded fiber length, and (b) as a function of fiber displace- ment, for various friction conditions [from Lawrence (30)]. 15 required to debond into consideration, and assumed that the debonding process is catastrophic as soon as the maximum shear stress exceeds the interfacial bonding strength. Instead they have derived the pull-out stress after complete debonding has occurred, in consideration of the decrease of the frictional stress due to Poisson contraction of the fiber under tensile stress. The initial pull-out stress is given by oi = A (1 — exp (-Bxe)) (Eq. 3) where x, is the embedded length of the fiber, A is a function of the normal compressive stress exerted by the matrix on the fiber across the interface and the elastic properties of the fiber and the matrix, and B is a function of the coefficient of friction between the fiber and matrix at the interface, and their elastic properties. Bowling and Groves (59) have developed a model for the debonding and pull-out of a ductile fiber embedded in a brittle matrix. They followed a theoretical treatment of the debonding stress proposed by Outwater and Murphy (60), based on an energy balance argument: ad = (8:77] + .2151? (Eq. 4) f f where Ef is the Young’s modulus of the fiber, rfis its radius, If is the fric- tional shear stress in the debonded region, 7 is the fiber-matrix interfacial energy and Id is debonded length of the fiber. According to their model, at a sufficiently large value of Id, the debonding stress exceeds the yield stress of the fiber and the relatively large plastic radial contraction of the 16 fiber reduces the frictional shear stress in the debonded region to zero. Therefore, they have concluded that debonding stress is independent of the embedded length and is determined by the yield stress and work-harden- ing rate of the fiber and roughness of debonded surfaces. A graphical interpretation and analysis of pull-out test results for duc- tile fibers embedded in a brittle matrix was proposed by Bartos (32), and Lawrence’s theory was extended to calculate the load-displacement curve during pull-out, the crack spacing and strength of aligned short fiber com- posites by Laws (46). Recently Hsueh (35) has provided rigorous solutions for stresses required to debond the fiber/matrix interface and to pull-out a fiber in a manner similar to shear-lag analysis incorporating shear defor- mation in the matrix. Several approaches have been used to analyze the interfacial shear stresses in push-out configurations (44,55,57,61-69), since the development of an indentation method for measuring frictional stresses at interface in ceramic matrix composite by Marshall. In Marshall’s calculations of slid- ing frictional stresses at the interfaces based on an energy balance analy- sis (55), it was assumed that the interfacial shear stress over the embedded length is uniform, and consequently the influence of Poisson expansion of the fiber on the frictional shear stresses at interfaces was neglected. Grande et al. (44) have analyzed the fully bonded fibers in push-out tests using finite-element method, providing a more accurate modelling of the localized point loading at the free fiber. A modified shear-lag analysis for the push-out tests has been proposed by Shetty (61). The Poisson expan- sion of the fibers under the compressive loads and the consequent increase of the normal stress aeross the interface leads to a nonlinear variation of the frictional shear stress along the embedded fiber length. Weihs et al. 17 (64,65) have attempted to incorporate Marshall’s energy balance analysis with Shetty’s consideration of Poisson expansion of the fiber. In addition, the importance of roughness of debonded surfaces to the frictional shear stress at the interface was considered by several research- ers (53,70,71) who have modelled its effect as an addition to the clamping pressure causing interfacial frictional shear stress. A detailed analysis of asperity interactions was presented by Carter et al. (70). Roughness is modelled as Hertzian contacts, leading to a sinusoidal modulation of the sliding pressure. Kerans et al. (53) included asperity pressure due to roughness of the interface in their treatment of clamping pressure, and have provided a discussion of abrasion during fiber sliding. Most recently Mackin et al. (71) developed a model considering an elastic asperity mis- match at the interface resulting in an asperity pressure at the interface and incorporating fractal models of interface roughness. 2-2. Fracture Toughness Measurements in Ceramic Matrix Compos- ites Ceramic matrix composites exhibit dramatic improvement in toughness compared with monolithic ceramics. Nevertheless, due to the difficulty in defining representative and universal characteristics of crack propagation arising from crack front deflection, crack wake delamination and crack bridging, it is hard to provide a meaningful technique to measure mechani- cal properties of fiber-reinforced ceramic matrix composite. A number of researchers have proposed various techniques of fracture toughness mea- surements in ceramic matrix composites (1-28,72-76). Single notched beam loaded in three point or four point bending is com- 18 monly used for fracture toughness measurement of ceramic matrix compos- ites (2-4,18,19). Such measurements are based on the linear elastic fracture mechanics, which assumes the same mechanical behavior over the whole sample and the occurrence of a very small process zone in the vicin- ity of the crack tip. The critical stress intensity factor (fracture toughness) for single notched beam in three point bending is given by (77) K, = Y—lnfl— (Eq. 5) where M max is applied bending moment at peak load, B is the thickness of the specimen, W is the width of the specimen, a is initial crack length, and Yis a geometry factor which is a function of a / W. However, as observed by Marshall et al. (78), debonding along the interface is accompanied with crack propagation normal to the fibers during failure of the ceramic matrix composites unidirectionally reinforced with fibers. Therefore, measure- ment of fracture toughness by flexural tests on the linear elastic fracture mechanics basis becomes invalid (78,79). In addition to this, the pull-out effect is not included in this analysis because the behavior after peak load is neglected. However, this technique may be applicable to ceramic matrix composites reinforced with particles, assuming isotropic behavior of the material due to uniform dispersion of small size particles. Several recent studies have dealt with J -integral method for analyzing the toughening due to crack bridging in ceramic matrix composites (5-8,10- 13,21-25,73,74). J-integral method developed by Rice (80) provides a solu- tion for considerable mathematical difficulties encountered in determina- tion of concentrated strain fields near crack tip in nonlinear elastic l9 materials. Path independent J -integral is defined by J = I(wdy— (T. 93)“) (Eq. 6) r Bx where F is a curve surrounding the crack tip, (0 is the strain-energy den- sity, T is the traction vector defined according to the outward normal along F, E is the displacement vector and ds is an element of arc length along 1". J is equal to the strain energy release rate as calculated from the usual continuum solution for small-scale yielding. Employing path independent J -integral in Dugdale-Barenblatt crack model (81) shown in Fig. 7, with dy=0 on F, Eq.6 can be written as J = jg‘o(8)d5 (Eq. 7) where 5, is the separation distance at the crack tip. Some researchers evaluated the fracture toughness due to crack bridging by intact fibers in ceramic matrix composites using J -integral method with same concept as Dugdale-Barenblatt crack model (Eq. 7) and obtained AGc = Jc .-.- ijg‘orumu. (Eq. 8) where AGc is the critical strain energy release rate, Vf is the area-fraction of reinforcement intercepted by the crack, o( u) is the normal stress acting on the reinforcement, and u" is the crack opening at the point where rein- forcement fails. However, if the crack opening distance in the crack bridg- ing zone is relatively large due to weak bonding, then dy in Eq. 6 can not be 20 in 5: SEPARATION DISTANCE gr {‘I I I I I I I I l Ezra-— v “II 31 t 0(5): RESTRAINING STRESS X (a) ‘— ‘— <— l 6. (b) Fig. 7 Dugdale-Barenblatt crack model (81). (a) Cohesive zone at crack tip with restraining stress dependent on separation distance; (b) force-displacement relation for atomic attraction in elastic brittle fracture. 21 considered as zero, and consequently Eq. 7 can not be obtained in this case. Therefore, J -integral method does not seem to be an appropriate method to evaluate the fracture toughness of ceramic matrix composite, especially with metallic fiber-reinforcement. However, taking only the result of this approach into account, Eq. 8 represents the strain energy release rate due to crack bridging by intact reinforcements. 2-3. Strength Analyses in Ceramic Matrix Composites Many ceramic matrix composites in which the failure strain of the matrix is less than that of the fiber experience cracking of the matrix when subjected to a uniform stress in the fiber direction. A critical stress at which the composite displays first evidence of the matrix cracking is called first matrix cracking stress. The first matrix cracking can cause damage to mechanical behavior of the composites. First of all, matrix cracking reduces the elastic modulus of the composite. Moreover, the intact fibers are exposed to oxidizing environment due to such cracks resulting in strength deterioration of the composite at elevated temperatures. There- fore, the first matrix cracking stress can be considered as a design stress in many applications of the composites at elevated temperatures. A number of micromechanics] models which relate the first matrix cracking stress to composite parameters have been developed (82-88). Aveston, Cooper and Kelly (ACK) (82) have analyzed the mechanism of matrix fracture in fiber-reinforced brittle matrix composites with friction- ally coupled interface. Their approach was based on an energy balance cri- terion under steady state cracking. According to this analysis, the matrix failure strain of the composite, smuc, will be enhanced only when 22 2 1/3 (Vi ) E’ (E 9) emu< 12mm 717; (E61531) q. where emu is the normal failure strain of the monolithic matrix, 1: is the fiber/matrix interface shear strength, 7m is the matrix fracture energy, r is the fiber radius, and Ef, EC and Em are the Young’s moduli of the fiber, com- posite and matrix, respectively. As long as the condition in Eq.9 is satis- fied, emuc increases with variation in composite parameters. On the other hand, as mentioned by Davidge (84), when the inequality in Eq. 9 is not obeyed, the matrix failure strain in the composite will be equal to the fail- ure strain of the monolithic matrix, i.e. emu=emuc. Further increase in strain results in matrix cracks normal to the fibers. Budianski, Hutchinson and Evans (BHE) (83) have treated both fric- tionally and adhesively coupled interfaces and studied the matrix cracking stress in a more rigorous fashion. They concluded that in the Coulomb friction-operative case, the larger strain mismatch would not lead to the higher matrix cracking stress, despite the larger frictional resistance avail- able. This is due to the fact that the same strain mismatches also lead to axial tensile stresses in the matrix which act to reduce the matrix cracking stress. Therefore, optimal strain mismatches exist for maximizing the matrix cracking strength. On the other hand, in the case of initially bonded fibers, the matrix cracking stress would increase with interface bonding toughness. More recently, Sutcu and Hillig (SH) model (86) also included the effect of both the adhesion and friction on the first matrix cracking stress, considering axial tensile stress to cause debonding. Each of these models assumed a steady-state matrix cracking and predicted dependence of the first matrix cracking stress on the interfacial shear 23 strength in brittle matrix composite. Singh (87) has recently compared the theoretical predictions of these models with experimentally observed values of matrix cracking stress in ceramic matrix composite. According to his experimental results, the matrix cracking is controlled by the critical strain for matrix cracking rather than the interfacial properties. He concluded that his results did not agree with the prediction of energy-based models such as ACK, BHE and SH models, but appeared to be in good agreement with a strain-based model suggested by Davidge (84). In addition to the first matrix cracking stress, the ultimate load-carry- ing capacity of the composite is of interest as well. Schwietert et al. (88) have presented a theory for computing the ultimate tensile strength of fiber-reinforced composites with multiple matrix cracking, considering the distribution of fiber breaks and their contribution to the diminished load- carrying capacity of the composite. They concluded that the strength of the composite depends sensitively on the interfacial shear stresses and on the fiber strength variability. 3. Experimental Procedure The composite specimens were fabricated by sandwiching technique (89) using as-received or acid-etched ribbons. The ribbon geometry for rein- forcement was selected to facilitate the ease of fabrication and theoretical modelling of the composite. Acid etching of the ribbon, in as-received or pre-oxidized conditions, was performed in order to medify the interfacial characteristics of the composite (42). Interface modification was also car- ried out by thermal cycling the composite specimens reinforced with as- received ribbon for various numbers of thermal cycles (41); the maximum temperature of the thermal cycles was also changed to achieve different thermal stress conditions. Pull-out tests of the specimens in various bond- ing conditions were carried out to evaluate the interfacial bonding strength. The fracture toughness of the composites was measured by three point bending test using single notched beam specimens, and by tensile testing pre-cracked specimens (22). 3-1. Materials The matrix material used in this study was soda-lime slide glass (73 wt% Si02 - 15 wt% Na20 - 4 wt% MgO - 7 wt% CaO - 1 wt% A1203 ). The dimensions of this slide glass was 76.2mm x 25.4mm x 1.1mm. Nichrome ribbon (80 wt% Ni - 20 wt% Cr) was chosen as the reinforcement. The Nichrome ribbons were 1.5 ~1.8mm wide and 80 um thick. Tensile proper- ties of a Nichrome ribbon with same heat treatment conditions as involved in the composite fabrication (450°C for 1 hour, and then 710°C for 3 hours) 24 25 was determined by Instron testing machine. The tensile stress-strain curve of the Nichrome ribbon used is shown in Fig. 8. The mechanical and physical properties of the matrix and ribbon used in current study are shown in Table 1. 3-2. Composite Fabrication Specimens were fabricated by sandwiching a series of aligned ribbons between the glass slides (89), as shown in Fig. 9. Prior to sandwiching the ribbons, the glass slides and Nichrome ribbons were cleaned with acetone and rinsed in distilled water. To avoid misalignment of the glass slides and ribbons during loading of the specimen into the furnace, a small amount of epoxy glue was applied to the corners of the glass slides. A pair of stainless steel plates (100mm x 40mm x 1.5mm), one as the base plate on which the specimen is placed inside the furnace, the other as cover plate on which the weight is applied, were utilized. A stainless steel block (100mm x 35mm x 10mm) (0.75 lbs) was used as a weight to achieve good contact between the matrix and the ribbon. The stainless steel plates were coated with BN spray in order to avoid sticking to the samples and to prevent their oxida- tion during fabrication. The composite fabrication involved two steps of heating: 450°C for 1 hour and 710°C for 3 hours. The low temperature step avoided gas trap- ping at the interface regions of the composites resulting from outgassing of the epoxy glue during fabrication. The high temperature step, which used a temperature of 10°C higher than the softening temperature of soda lime glass, allowed the glass to flow and cause adherence to the reinforcements. The weight (stainless steel block), and the top stainless steel plate were STRESS (MPa) 26 unit: mm R:6.4 \ i 5 10 s ‘*—'127'—*fi 700— 600- 500- 400- 300- 200- 100- 0 l I I l I I I l T j 01) 01 ()2 0&3 OA- ()5 STRAIN (mm/mm) fig. 8 Typical stress-strain curve of the Nichrome ribbon obtained from tensile test. 27 Table 1: Physical and mechanical properties of constituents of the composite Matrix Ribbon soda lime Material slide glass Nichrome Thermal Expansion Coeffi- 9.2 14.3 cient (10'6/°C) , (25 °C ~ 500 °C) Young’s Modulus (GPa) 72 213 Yield Strength (MPa) 305 Tensile Strength (MPa) 700 Softening 'Ibmperature (°C) 700 28 I ::':11211;222:2335:2:2;2;2;t~:°:~:-:°:'Rot-:0}:-:-:tot-9:40:09??? ‘ WBIth .,_. A_.,l,-.; .' ‘ . r. ICU"; .4.- ‘-I-‘.’-1a;~ . utvn :5. ~.. -' ‘7‘ Sta|nless Steel p'ate .-:::_I.""TT"““TT'Tmimff'” ”tr-"mm- ~ nah-er - l~" ‘ ”W!- “W _ BN t. g . l. ~1- r W coa 'n Sllde glass—— _ — ——--N' h 'bb IC rome II on .M'IJJ M...“ ”:5“ ;‘i u “M ‘7‘!— (a) Fig. 9 The schematics illustrating sandwiching of a series of aligned ribbons between the glass slides: a) general set-up, b) composite from which pull-out test samples were cut out, and c) composite from which fracture toughness test samples were cut out. 29 removed after cooling to 500°C. The composites were annealed at 500°C for 2 hours, then cooled to room temperature in the furnace. Entire com- posite fabrication was carried out in ambient atmosphere. Pull-out specimens (4mm x 10mm x 15mm) and fracture toughness test specimens (5mm wide and 6mm thick) were cut out from the fabricated composite stocks using a low speed diamond saw. The schematics of typical pull-out and pre-cracked tensile specimens are shown in Figs. 10 and 11 respectively. The composite specimens cut to proper dimensions were mechanically polished with a series of abrasive grit papers ranging from 240 grit to 600 grit. 3-3. Interface Modification 3-3-1. Thermal Cycling Technique Thermal cycling method used in this study involved placing the specimens on a platform in the center of a standard laboratory furnace (Lindberg) at various temperatures for 3 hours, followed by cooling in still air and hold- ing at room temperature for 3 hours. The maximum temperatures used for thermal cycling were 500, 550, 600, and 650°C. These temperatures lie between the strain temperature (480°C) and softening temperature (700°C) of soda lime glass. These thermally cycled specimens are referred to as 500-TC, 550-TC, 600-TC, and 650-TC specimens in this study. The number of thermal cycles chosen in this study were 3, 5, and 7 cycles for each maximum cyclic temperature. 30 Y (transverse) j—v X (longitudinal) Z (lateral) MATRIX f l RIBBONI Fig. 10 The schematic of typical pull-out specimen. 31 a =1 ~1.2mm Id—B——> (a) B = 6.0mm W=5.0mm w = 1.8mm t = 80 um i) I ‘ I6 is variable. nbbon ' w - le pro-crack -‘ ' f ,_ =7 y mTjr I (b) Fig. 11 The schematic of typical fracture toughness test specimens: a) single notched specimen for flexure test, and b) pre- cracked specimens for tensile test. 32 3-3-2. Acid Etching Technique Prior to the composite fabrication, the surfaces of Nichrome ribbons were etched in order to modify the interfacial characteristics of the com- posites. The ribbons in both as-received and pre-oxidized conditions were acid etched. Pro-oxidizing of the ribbon involved holding the Nichrome rib- bons (cleaned with acetone and rinsed with the distilled water) at 650°C for 1 hour. The oxide layer produced by such a step reduced the intensity of acid etching on the ribbon surfaces. The Nichrome ribbons, in both con- ditions, were etched with a solution of 20% HNO3 and 80% HCl for various lengths of time: 5, 15, 30, 60 and 90 seconds. 3-4. Mechanical 'lbsting Pull-out tests were carried out on the composite specimens with various interfacial bonding conditions to evaluate the interfacial bonding charac- teristics. These tests were conducted on a computer controlled Instron testing machine (model#4206), with a specially designed self-aligned fix- ture. The crosshead speed used was 0.5mm/min. For effective tensile grip- ping, the glass part of the specimen was held by glass-fiber reinforced epoxy tabs which was glued to the fixture (90). The glass fiber-reinforced epoxy tabs were used to prevent the cracking of glass matrix during tight- ening of the sample to the fixture. In addition, a dab of glue was also applied to the glass part of the specimen, before putting it on the fixture, to avoid slippage during pull-out test. The schematic of the experimental set- up for the tensile test is shown in Fig. 12. 33 crosshead > load cell U 33"}? 3——> upper grip \J UUV self-aligning fixture lower grip 4 .I I I' 4 .I .I .I I l .I .I .I .I I I' 4 .I .I .l I I’ a .I I I (a) Fig. 12 A schematic of the experimental set-up for the tensile test: a) general set-up, b) self-aligning fixture. The aligning rails were removed after positioning the specimen in the testing machine. Inside of fixture covers glass reinforced epoxy supporting 1 rod Fig. 12 (continued) + screw hole --> pull-out sample —> aligning rail i—r screw i 35 Fracture toughness behavior of the composite was evaluated by both flexure and tension tests on the same specimen. The tension tests were carried out on pre-cracked specimens resulting from the flexure tests. The length of the specimens was varied from 20mm to 30mm. For flexure test, the composite specimens (5mm wide and 6mm thick) were initially notched by a slow speed diamond saw at the center of the specimen. Since diamond saw provides a blunt notch, a razor blade was used to produce a sharper i notch at the base of the blunt notch. The resulting depths of the notches were measured after tension tests. Those values ranged between 1mm- I 1.2mm. These single notched specimens were loaded on a three point bending fixture with a 18mm span in an Instron testing machine. The crosshead speed was 0.2mm/min. Bending of the composite specimen was continued until the matrix cracked, as indicated by a drop in load. Due to such a procedure, the specimen contained a continuous matrix crack around the Nichrome ribbon. A sketch of a pre-cracked specimen is shown in Fig. 11. Using values of the peak load obtained from three point bending tests, ch was calculated using linear elastic fracture mechanics (7 7). Ten- sion tests were carried out with these pre-cracked specimens, in order to study the effect of crack bridging on the fracture toughness of the compos- ite (22). A tensile testing machine (Instron) was used in these tests with a specially designed self-aligning fixture (Fig.12). The crosshead speed was 0.5mm/min. The pre-cracked specimens were placed in the self-aligning fixture. The ends of the specimens (to be gripped) were held by glass-fiber reinforced epoxy tabs which were already glued to the fixture. In addition, a dab of glue was also applied to the ends of the specimen (to be gripped) before putting it on the fixture. The stress-displacement curves were 36 obtained using the Instron testing machine at a crosshead speed of 0.2mm/ min. The areas under the curves were measured, and the values of the fracture toughness due to crack bridging were calculated by using Eq. 8. 3-5. Micrographic Observation For the interfacial studies using SEM, the specimens were cut perpen- dicular or parallel to the specimen length with a low speed diamond saw. The cut samples were mounted in lucite and mechanically polished with a series of abrasive grit papers ranging from 240 grit to 600 grit followed by polishing on cloth using 5pm to 0.3um size alumina powder abrasives. The polished specimens were etched with a solution of 20% HNO3 and 80% HCl for 20 seconds. Fractographic observations were made on the specimens fractured during tensile loading, using the Olympus stereo optical micro- scope equipped with two fiber-optic transmitted light sources. The direc- tions of light sources were changed until a interference fringe pattern representing the crack path could be observed. 4. Results and Discussion 4-1. Theoretical and Experimental Analysis of Debonding Process of Ductile Ribbon Embedded in Brittle Matrix 4-1-1. Theoretical Modelling Several analytical models of debonding process during fiber pull-out in fiber-reinforced ceramic matrix composites have been developed (29-36). Among these, the model suggested by Lawrence (30) was adopted in this study for elastic behavior of the sample shown in Fig. 8. The ribbon geom- etry of the reinforcement was used in this study so as to simplify the mod- elling and to achieve easy fabrication of composites. The distribution of shear stress (1:) along the interface at a point with a distance x from the embedded end of the ribbon due to applied load (P) is given by (30) (Eq. 10) 1: _ Pa coshax ' 2(w+t) (Sinhale) where w and t are width and thickness of the ribbon respectively, I, is embedded length of the ribbon, and a is a shear-lag parameter dependent on elastic properties of the matrix and the ribbon and on their dimensions (31). The maximum shear (“rm”) which is obtained at the boundary where the ribbon enters the matrix is Pa Tmax = 2(w+t) cothale (Eq. 11) 37 38 Since the values of coth (11,, approach unity for larger values of le, tmax can be considered to be independent of I, for large values of le. 4-1-1-1. Load Required to Initiate Debonding In ceramic matrix composite reinforced with metallic ribbon, the bound- ary where the ribbon enters the matrix is affected by plastic deformation of the unembedded region of the ribbon. Therefore, the shear stress due to plastic deformation of the ribbon as well as the shear stress due to shear- lag analysis acts at the interface along the boundary between free and embedded areas of the ribbon. These shear stresses are shown in Fig. 13. For simplification of this modelling, this can be considered as a 2-dimen- sional model because the thickness of the ribbon is negligible compared to its width and length. Accordingly, the sketches in Fig. 13 illustrate long transverse faces of the specimens. The amount of shear stress that can contribute to debonding (1:8) equals to the resolved shear stress available in excess to that required to initiate plastic deformation, and Is is assumed to act along a direction at 45° to the pull-out direction. This assumption can be explained by Hall-Patch model for deformation of polycrystals (91). o P-wtoo = —— (Eq. 12) 1-1 -1: -O—° - -2 .2— 2wt 8 rss yrss where 1,8, is resolved shear stress due to applied normal stress, 1:3,,“ is resolved shear stress required for yielding of the reinforcement, o is applied normal stress, and do is yield stress. 18 is combined with In,” to obtain the resultant shear stress (It) at the interface along the boundary between free and embedded areas of the ribbon, so that 39 Fig.13 Shear stresses due to tensile loading at edge of the embedded area Irma, : shear stress obtained by modified shear lag model (30), t, : resolved shear stress available in excess to that required to initiate plastic deformation of the ribbon by slip, and 1. : resultant shear stress due to 1m” and 1,] Note: 1, is at 45° to the direction of P. 40 1:t = «ltrznax + Jz-tmaxts + If (Eq' 13) When resultant shear stress along the boundary between free and embed- ded areas of the ribbon (1,) exceeds the interfacial shear strength of the composite (ti), interfacial debonding initiates. Therefore, the interfacial shear strength of the composite can be written as ri=frxd )2+J§(rd max max ) (13H (1,852 (Eq. 14) where superscript ‘d’ corresponds to debonding condition. Accordingly, Pd-wto Pdor a max = WcOthale and I: = W (Eq. 15) where Pd is the load required to initiate interfacial debonding of the com- posite. 4-1-1-2. Load Required for Continuation of Debonding After initiation of interfacial debonding, static frictional force (Pf) arises at the debonded interface due to residual gripping stress (on). This residual gripping stress is due to thermal expansion coefficient mismatch between the ribbon and the matrix (31). Therefore, in order to continue to debond, the applied load (PI) should overwhelm this static frictional force as well as the load required to debond at the bonded/debonded boundary (Fig. 14). 41 Fig. 14 Load(P Jrequired for continuation of debonding: P > P, + P,f where P,d 18 force required to debonding, and P, x" 18 the frictional force between debonded region (shaded area) of ribbon with the matrix. 42 The load required to continue debonding at the bonded/debonded boundary (P,d) can be determined from (Eqs. 14 and 15) under the assumption that interfacial shear strength (12,-) is uniform through the whole interface of the composite. At the location marked Q in Fig. 14, the debonded length is (l,- x), and the interfacial shear strength is 2 2 r, = Jeff) + Jé (cf) (1?...) + (1?...) (Eq. 16) where Pda P? -wto = x d = ____° . x 2 (w+t) cothorx, and 1: ’ (Eq 17) The static frictional force (P,f) at the debonded interface can be determined in the following manner. Assuming coefficient of friction (u) to be a con- stant throughout the debonded interface, frictional force change (dP,f) with infinitesimal change in distance (dx) at Q (Fig. 15) is dPJft = 2 (w+t)uondx (Eq. 18) However, under external tensile loading, contraction of thickness of the ribbon due to elastic and plastic deformation reduces the gripping stress from the matrix, resulting in decreased frictional force. Reductions in the gripping stress from the matrix result from Poisson’s effect due to elastic deformation by he, (31) and from slip due to plastic deformation by as, where k is a constant depending on the elastic properties of the constitu- ents of the composite, a, is normal stress at Q in the direction of the ribbon pull-out, and a, is a component of r, mentioned in previous section in the 43 W//////////////% Fig.15 Static frictional force at the debonded interface : contraction of thickness of the ribbon due to elastic and plastic defama- tion reduces the gripping stress from the matrix. 44 thickness direction of the ribbon. Under such conditions Eq. 18 can be rewritten as de = 2(w+t)u(on-kor—os)dx (Eq. 19) a, in this equation can be replaced by 0'0 since the ribbon would have yielded under these conditions. The static frictional force at Q (P,f) can be obtained by substitution of o, and o", and by integration of Eq. 19 from I, through 1:: x (w'I't) 1: pl; =2(w+t)uL.(on-koo)dx- J2wt u ,(PZ—wtcowx (19,20) In order to continue debonding, the applied load (P1) should become 19,310;l +P§ (Eq. 21) Further plastic deformation of the ribbon after debonding is neglected in this analysis. 4-1-2. Experimental Analysis A typical load-displacement curve obtained in pull-out test of the Nichrome ribbon embedded in soda-lime glass matrix is given in Fig. 16. In this test, the load linearly increases up to the point ‘A’, corresponding to the yield point of the ribbon. The free (unconstrained) area of the ribbon starts to deform plastically at point ‘A’, and work hardens on further defor- mation. This smooth work hardening curve is interrupted by small load STRESS(MPa) 45 700- . C Pull—out. starts 600- _ Debonding . starts 500- 400- . - A 33:338. 2.1.2:: 300- soc-JI 100- 0 r I I I I I I l I I I [ fi 0 1 2 3 44 5 6 '7 DISPLACEMENT(mm) Fig.16 Typical stress-displacement curve obtained in pull-out tests of Nichrome ribbon embedded in soda lime glass matrix. 46 drops, and the first load drop ‘B’ is associated with initial debonding at the ribbon-matrix interface. As speculated by Morten et al. (90), the debonding front actually propagates in small jumps rather than continuously, accom- panied by the small load drops on the load-displacement curve. After ini- tial debonding, the load continuously increases as debonding progresses until the load reaches a peak point ‘C’. At point ‘0’, complete debonding occurs along the full embedded length, and at the same time the ribbon pull-out initiates. The sudden load drop at this initiation of pull-out of the ribbon was not observed in this experiment. The explanation for this fact will be discussed in detail later. After initiation of pull-out, the load con- tinuously decreases to zero until the ribbon is completely pulled out from the matrix. SEM micrographs of the ribbon/matrix interface are shown in Fig. 17. A relatively thin oxide layer that can be seen at the interface facilitates chemical bonding between glass matrix and ribbon. The interface of this composite contains partially debonded regions due to the considerable mis- match in thermal expansion coefficients of the ribbon and the matrix. The crack was observed along the weak interface between the oxide layer and the Nichrome-metal. In this composite, the ribbon experiences tensile ' stresses in the lateral, longitudinal, and transverse directions due to the larger thermal expansion coefficient of the ribbon than that of the matrix. Therefore, once debonding occurs, the ribbon contracts not only in the lat- eral direction but also in both longitudinal and transverse directions. Con- sidering dimensions of the ribbon, displacements due to longitudinal and transverse contraction of the ribbon are much larger than that due to lat- eral contraction. On the other hand, tensile loading during pull-out test gives rise to longitudinal plastic extension of the ribbon. As a result, longi- 47 MATRIX Fig.17 SEM micrograph of the partially debonded interface between the Nichrome ribbon and the soda lime glass matrix 48 tudinal contraction of the ribbon due to release of the residual tensile stress by debonding vanishes and only plastic extension of the ribbon exists in the longitudinal direction. ' Consequently, transverse contraction and longitudinal plastic extension of the ribbon, coupled with the micro- scale roughness of the debonded surfaces, result in the mechanical keying of the debonded surfaces providing compressive stresses at the interface, as schematically shown in Fig. 18. The SEM micrograph of the interface between the ribbon and the matrix after debonding progresses to some degree is given in Fig. 19. To confirm progress of the debonding process, loading had been interrupted just before the ribbon can pull out from the matrix. The shear failure at the interface can be observed at the deep embedded location (M). It could have resulted from the plastic contraction in width of the ribbon as well as from the distribution of the shear stresses along the interface due to tensile loading. Although further plastic defor- mation of the ribbon after debonding was not considered in this analysis, this behavior was also indicated by increased crack opening of initially debonded location (N) in this figure. The interfacial shear strength (1:5) of this composite as a function of embedded ribbon length is shown in Fig. 20. These values of the interfa- cial shear strength of this composite were obtained by equations (14) and (15), using the results of load-displacement curves obtained experimen- tally. The interfacial shear strength of this composite was evaluated to be 97.9 :l: 15.0 MPa. Plot of loads required to continue debonding as a function of debonded length, obtained on the basis of section 4-1-1-2, is given in Fig. 21. The load required for continuation of debonding (P,) is the sum of the load required to debond at the bonded/debonded boundary (P,d) and static frictional force (Pf). P,d gradually decreases with increasing debonded 49 MATRIX > SURFACES \/\/\/\/ <-—- RIBBON 00 W \ \ \ Fig.18 Mechanical keying of debonded surfaces produced by rough- ness of debonded surfaces, transverse contraction, and longi- tudinal plastic extension of the ribbon due to tensile loading after debonding. 50 MATRIX RIBBON RIBBON MATRIX Fig.19 SEM micrograph of the interface between the Nichrome rib- bon and soda lime glass matrix after debonding progresses to some degree 51 ,\ 200- ‘U .. Q. E q :3 . S a z 150‘ a: . 01 .. 9* U) ' 0 U " ' AV. : 97.9MPa E 100- . . Q - o 5 ~ - - a] . 2] . E3 50- < 1 E4 . D: m a: 5‘ . E O I l I l I 2 3 4 EMBEDDED LENGTH (mm) Fig.20 Interfacial shear stresses for debonding various embedded lengths of the Nichrome ribbon 52 DEBONDED LENGTH (mm) Fig.21 Loads required to continue debonding as a function of deb- onded length : P, = 1D," + P,’ 53 length (le-x) at a very low rate in the beginning. However, when the bonded length(x) remains very short, P,d rapidly drops. This phenomena is attributable to the behavior of function coth our in the expression for t,d, and the contribution of plastic deformation of the ribbon (1:8,,d) (Eq. 17 ). Especially, at short bonded length, since P,d is not large enough to yield the ribbon, the contribution of plastic deformation of the ribbon does not affect the plot of P,d. P,d eventually becomes zero when complete debond- ing is accomplished. On the other hand, the static frictional force (P,f) monotonically increases with debonded length (le-x), with a rapid increase at small :5 due to the absence of plastic deformation of the ribbon. Since the plastic deformation of debonded ribbon is neglected in this analysis, the values of P,f obtained will be overestimations. Therefore, the sum of the above two components which represents loads required for continua- tion of debonding (P,) must also be overestimated in this analysis, although the difference of maximum peak loads between this analysis (94.5N) and actual result (92.0N) was not remarkable. As shown in Fig. 21, loads required for continuation of debonding (P,) steadily increases to a maximum and then drops. Therefore, the debonding process in this case is catastrophic at the maximum required debond load. In general, it is said that if catastrophic debonding occurs, the load suddenly drops in load-dis- placement curve (34). However, this sudden load drop was not observed in load-displacement curve in this experiment. It can be explained by the roughness of the debonding surfaces of the composite which influences the sliding frictional force during the pull-out of the ribbon from the matrix. The roughness of the debonded surfaces that exist in this composite (Fig. 19) may contribute to this behavior. The effects of embedded length on the normal stress acting on the rib- | .1 ' LI -3' I :L 54 bon when debonding starts (6;), and normal stress when debonding is com- pleted (Of) are shown in Fig. 22. These stresses were obtained by dividing the corresponding loads by the ribbon cross-sectional area. The experi- mental results indicate that bath a,- and of remain constant for embedded lengths in excess of a critical length. The result of o,- agrees well with the theoretical analysis by Hseuh (35), which dealt with elastic fiber/elastic matrix system. This good agreement may be due to the fact that the term accounting for plastic deformation of ribbon (Eq. 12) is independent of the embedded ribbon length. On the other hand, contraction in thickness due to further plastic deformation of the ribbon after debonding gives clearance to debonded interface, resulting in no static frictional farce. Therefore, no matter how long the embedded length may be, only certain newly debonded length undergoes static frictional farce. Since ofis mainly affected by ini- tial debonding stress (6;) and static frictional force at the debonded inter- face as illustrated in Fig. 21, it reaches a constant value for embedded length in excess of a critical value. 55 700! 03 O O l J 0‘ O O l 4:. O O l NORMAL STRESS (MPa) O l I I l I I I I I 0 1 2 3 4 EMBEDDED LENGTH(mm) (”J Fig.22 Effect of embedded length on the normal stress on the ribbon when debonding starts (6;), and normal stress when debond- ing is completed (6,) in soda lime slide glass reinforced with Nichrome ribbon. 56 4-2. Interface Modifications 4-2-1. Thermal Cycling Technique Typical stress-displacement curves of as-fabricated, and 650-TC speci- mens that have undergone 3, 5 and 7 cycles given in Fig. 23 illustrate that these thermal cycles significantly affect the mechanical behavior of the composite. However, stress-displacement curves of 500-TC specimens are about the same as that of the as-fabricated specimens. This behavior indi- cates that the thermal stresses developed in 500-TC specimens might not be high enough to cause microcracks at the interface after 5 thermal cycles. As shown in Fig. 24, in 650-TC specimens microcracks are generated at the interface after 3 thermal cycles and they link to form an almost completely connected crack after 5 thermal cycles. These SEM findings account for the observed stress-displacement behavior of the composite due to thermal cycling. In addition, a variation in stress-displacement curves for TC spec- imen, with various maximum cyclic temperatures, is apparent in Fig. 25. These curves were obtained after 5 thermal cycles up to each maximum cyclic temperature. Although a maximum temperature of 500°C does not cause microcracks at the interface after 5 thermal cycles, microcracks do accumulate at the interface after 5 thermal cycles with increasing maxi- mum cyclic temperatures. The experimentally measured values of interfa- cial bonding strength and frictional shear stress are summarized in Tables 2and3. 57 7001 600- ”a? 0.. 500- 5 Cf) .l E3 400- 3 cycles 0:. E. d U) ._‘1 300- <2 E c- E .. O 200 Z . 100 0I ' I ' H—Fx I I I T I 1 2 3 4 5 6 '7 DISPLACEMENT(rnm) Fig.23 Typical stress-displacement curves of as-fabricated, and 650- TC specimens after various numbers of thermal cycles. 58 (a) Fig.24 SEM micrographs at interfaces of (a) as-fabricated speci- mens, (b) 650-TC specimen after 3 cycles, and (c) 650-TC specimen after 5 cycles. 59 (b) Fig.24 (continued) 60 (c) Fig.24 (continued) 61 . AS-FAB 500°C 0" O O l A O O l NOMINAL STRESS(MPa) - C _ I - INSPLACEMENThnnfl Fig.25 Stress-displacement curves of specimens that have under- gone 5 thermal cycles with variaus maximum temperatures. 62 ous numbers of thermal cycles Table 2: Measured interfacial properties of 650-TC specimens after vari- thermal cycling conditions interfacial properties as-fab 3 cycles 5 cycles - 7 cycles interfacial bonding 97.9 74,9 20.7 7.61 strength 1218.2 110.2 (MPa) i156 frictional shear 9.24 3,33 2.62 1.05 stress 1 2.16 fl 13 $1.07 :l:0.44 (MPa) ' 63 Table 3: Measured interfacial properties of thermal cycled specimens up to various maximum temperatures after 5 thermal cycles thermal cycling conditions interfacial properties as-fab 500°C 550°C 600°C 650°C interfacial bonding 97.9 98.1 83.7 45.2 20.7 strength i182 :l:7.8 :84 110.2 (MPa) frictional shear 9.24 10.26 5.05 4.90 2.62 stress :l:2.16 i248 i218 $1.07 (MPa) 64 4-2-1-1. Effects of thermal Cycling on The Interfacial Bonding Strength and Frictional Shear Stress The interfacial bonding strengths of 650-TC specimens that underwent various numbers of thermal cycles are given in Fig. 26. These were obtained from modified shear-lag analysis, taking plastic behavior of the metallic ribbon into account as mentioned in sec. 4-1-1 (Eqs. 14 and 15). The interfacial bonding strength decreases with increasing number of cycles and drops significantly between 3 cycles and 5 cycles. As mentioned earlier in 650-TC specimens, microcracks were generated after 3 cycles, and most of them linked together after 5 cycles, and resulted in a com- pletely connected crack at the interface after 7 cycles. As a matter of fact, most of the specimens that have undergone 7 thermal cycles showed com- pletely debonded interfaces. The value of the interfacial bonding strength for this condition shown in Fig. 26 was obtained from only one specimen that contained some bonded interface. The effect of maximum cyclic tem- perature on the interfacial bonding strength after 5 thermal cycles is pre- sented in Fig. 27. Rapid decrease in interfacial bonding strength occurs between 500°C and 600°C. In addition, the frictional shear stresses on the debonded surfaces of 650-TC specimens after various numbers of thermal cycles, and after 5 thermal cycles at various maximum temperatures are presented in Tables 2 and 3 respectively. These values of the frictional shear stress were calcu- lated under the assumption that frictional shear stress is constant over the debonded region, and were obtained from dividing the initial pull-out load by total embedded area. As can be noticed from Tables 2 and 3, the fric- 65 A 140- (U m d E 120 :1: " __ 5‘ .. E 0.1 E“ q U) U 80- — E . O z 60- O m I j 40- 8 I E Q: 20" 1 E13 2 ‘ o 0 I I l l l I l I I 0 1 2 3 4 5 6 7 8 NUMBER OF CYCLES Fig. 26 Interfacial bonding strengths of 650-TC specimens after vari- ous numbers of thermal cycles. 66 A 140- (D [L . 5 120 E . ‘i’ LzD as Eq - m I 0 so- —— g . a z 60- o m I O :3 4a- 5 . <2 fi 20- I m , E“ III E O I I I I I I I l I l I I I I a 100 200 300 400 500 600 700 MAXIMUM CYCLIC TEMPERATURE( C) Fig. 27 The effect of maximum cyclic temperature on the interfacial bonding strength after 5 thermal cycles. 67 tional shear stress on the debonded surfaces decreases with increasing number of thermal cycles, and with increasing maximum cyclic tempera- ture. It is believed that continual rubbing between the debonded surfaces with continued thermal cycling, or higher mismatch strain due to differ- ence in thermal expansion coefficients with increasing maximum cyclic temperature, causes smoothening of the debonded surfaces. As a result, the frictional shear stress becomes lower due to less mechanical keying effect of debonded surfaces. 4-2-1-2. Effects of Thermal Cycling on The Complete Debonding Stress The complete debonding stress is defined as the normal stress on the ribbon required to completely debond along the matrix-ribbon interface and is given by the maximum stress in the stress-displacement curves obtained in pull-out tests. The complete debonding stresses as a function of the embedded ribbon length in 650-TC specimens after various numbers of cycles is given in Fig. 28. The results illustrate that the complete deb- onding stress decreases with increasing number of thermal cycles in 650- TC specimens. After initiation of debonding, the frictional shear stress, which must be overcome to continue to debond, exists at the debonded interface due to roughness of the debonded surfaces and also due to ther- mal expansion mismatch between components. Therefore, the difference between the values of the complete debonding stress and the initial deb- onding stress, which represents the normal stress on the ribbon when deb- onding starts, can be considered to be such a frictional shear stress. The magnitude of frictional shear stress can be accounted for by the roughness of the debonded surfaces. As shown in Fig. 29, debonding due to thermal COMPLETE DEBONDING STRESS(MPa) 68 AS—F‘AB. (o) \1 O O I 600- ‘ fi>3 CYCLES (o) a: 500_ ' 5 CYCLES (2x) 400- 300- x x ,/*” 200- 100- . ' . 7 CYCLES (v) O l I I I I l I l I l O l 2 3 4 5 EMBEDDED LENGTH(mm) Fig. 28 The complete debonding stresses as a function of the embed- ded ribbon length after various numbers of thermal cycles of 650-TC specimens. 69 \z »' ~ 2* .‘ ' y “"3 * . - jf'?I\E¢-. I. :1 .>*“‘ 5"{W , [-5 ‘Y . u k | @L.fr.tx’~7: ‘ ’ w .. ~e"r kn-H: ‘L‘r’q-m (a) Fig. 29 SEM micrographs at debonded interface (a) due to ther- mal stress, and (b) due to shear stress during pull-out tests. 70 (b) Fig.29 (continued) 71 stress generates smooth surfaces [Fig 29(a)], while debonding due to shear stress realized in pull-out tests of embedded ribbons generate rough deb- onded surfaces [Fig. 29(b)]. Smoother debonded surfaces result in lower frictional shear stress, as mentioned in the previous section. In the case of as-fabricated specimens, after debonding initiates, stresses required for continuation of debonding steadily increase up to complete debonding stress due to higher frictional shear stress at the debonded interface resulting from rough interfacial cracking. However, in 650-TC specimens that have undergone 3 and 5 cycles, the frictional shear stress at the deb- onded interface is too low to contribute to stresses required for continua- tion of debonding. As a result, catastrophic debonding was observed in stress-displacement curves of these specimens. In brief, the complete deb- onding stress has two contributors; one is the initial debondingstress, the other is the frictional shear stress. Since both contributors decrease with increasing number of cycles, the complete debonding stress decreases with increasing number of thermal cycles. The effect of temperature on the complete debonding stress can be seen in Fig. 30. The complete debonding stresses of 650-TC specimens after 5 cycles are considerably lower, com- pared to those of as-fabricated ones, because both initial debonding stress and frictional shear stress decrease with increasing maximum cyclic tem- perature. However, 5 cycles of 500-TC do not affect the complete debonding stress or the stress-displacement curve and the interfacial bonding strength of the as-fabricated composite. COMPLETE DEBONDING STRESS(MPa) 72 700— D500°C (0) . AS-FAB.(0) 600- 550°C (7) 500- 650°C (x) 400- 300- x x 2004 * .. 3K 100- 0 l I I l I I I l I 1 o 1 2 3 4 5 EMBEDDED LENGTH(mm) Fig. 30 The effect of maximum cyclic temperature on the complete debonding stress 73 4-2-2. Acid Etching of The Ribbon The surfaces of the as-received and pre-oxidized ribbons after acid etching for a given time are shown in Fig. 31. The shapes of the dimples formed by acid etching the as-received and pre-oxidized ribbons are significantly dif- ferent from each other. Since the oxide layer protects the metallic ribbon p from severe attack by acid in the case of etching the pre-oxidized ribbon, the resultant density and size of dimples formed on the metallic ribbon surface were significantly reduced as compared to those in the case of etch- ing the as-received ribbons, as can be observed later in the micrographs of metallic ribbon surfaces after ribbon pull-out. The effect of ribbon surface treatment on stress-displacement curves obtained from pull-out tests of the N ichrome ribbon embedded soda-lime glass matrix sample is illustrated in Fig. 32. The pull-out stress, defined as the normal stress on the ribbon required to initiate ribbon pull-out from the matrix, significantly changed with ribbon surface treatment. Preoxida- tion at 650°C for 1hr did not affect the debonding and pull-out processes in this composite. This was due to the fact that the oxide layer was also cre- ated on the as-received ribbon during composite fabrication and the deb- onding crack predominantly propagates along the interface between metal and oxide layer in the pull—out test as can be observed in Fig. 33. On the other hand, acid etching of the Nichrome ribbon for 60 seconds led to a large improvement in the pull-out stress of the composite. Acid etching for the same time after pre-oxidization of the ribbon at 650°C for 1 hour, how- ever, showed only a moderate improvement in the pull-out stress. SEM micrographs of the ribbon and the matrix surfaces after ribbon is 74 (a) Fig. 31 Ribbon surfaces roughened by acid etching of (a) as-received and (b) pro-oxidized ribbons for 90 seconds 75 (b) Fig. 31 (continued) 76 700- C . A: as—received 600- D A B: pre—oxidized . M C: etching of as—received ’5 // ribbon for GOsecs. 0.. 500- , D: etching of pre-oxidized E , B ribbon for 60secs. m . g 400- / D: E. . m .4 300- ‘ <2 E q 2 .1 C) 200 Z . 100- 0 ' I E I T I ' I '— I I I ' I ' I O 1 2 3 4 5 -6 7 8 DISPLACEMENT(mm) Fig. 32 The effect of ribbon surface treatment on stress-displacement curves obtained from pull-out tests. 77 Fig. 33 Debonded surface in which debonding crack has propagated along the interface between metal and oxide layer. Oxide layer shown is stuck to the matrix after complete debonding. (A:matrix, Bzoxide layer) 78 completely pulled-out from the matrix for each case of surface roughening treatment are shown in Fig. 34. The abrasion process of the jagged inter- face of the composite with the as-received ribbon roughened by acid etching for 60 seconds is illustrated in Fig. 34(a). In the composite with the preox- idized ribbon roughened by acid etching for 60 seconds, the portions of glass which fill the dimples on the ribbon at the jagged interface failed by shear and stuck to the ribbon during ribbon pull-out as can be seen in Fig. 34(b). Fig. 35 provides schematic illustrations of the mechanisms of ribbon pull-out processes of the composites for the different ribbon surface treat- ments. During composite fabrication in air, an oxide layer is formed on the riben surface already roughened by acid etching [Fig. 35(a)]. Dimples formed on the ribbon surface by etching, are filled up with the glass matrix. During loading, debonding crack front propagates under shear in the same manner as mentioned earlier. When debonding along the inter- face is complete, the ribbon is not pulled-out right away because of mechanical interlocking at the jagged interface. However, once the higher stress required to overcome the interlocking at the interface is attained, the jagged interface starts to move. This moving interface scratches the metallic ribbon surface until the interface wears out, and flattens. This mechanically interlocked interface, which exhibits features of the abrasion process of the jagged interface, results in considerable increase in the pull- out stress as compared to that of the composite reinforced with the as- received ribbon. On the other hand, for the case of pre-oxidized ribbons, after a thin oxide layer has been created on the smooth as-received ribbon surface, the ribbon is roughened by acid etching. Since the oxide layer alleviates acid attack on the metallic ribbon surface, the effect of mechani- cal interlocking at the jagged interface is considerably reduced under the 79 RIBBON (a) Fig. 34 SEM micrographs of the ribbon and the matrix surfaces after ribbon is completely pulled-out from the matrix for the com- posite with: a) the ribbon roughened by acid etching for 60 seconds, and b) the pre-oxidized ribbon roughened by acid etching for 60 seconds. Note that the abrasive process has left striations in both the matrix and ribbon in a), and por- tions of glass stuck to the ribbon appear as bright specks in b). 80 MATRIX Fig. 34 (continued) 81 RIBBON (b) Fig. 34 (continued) 82 MATRIX Fig. 34 (continued) 83 After acid etching, ‘—\./ v v flbbon After composite fabrication, During pull-out test, -—-\,—-\7 1m? , arr—— debonding crack During ribbon pull-out, l malrIX ~ _> (a) Fig. 35 Schematic profile of the interface after ribbon surface treat- ment, composite fabrication, and during pull-out test for the composite with: a) as-received ribbon roughened by etching, b) pre-oxidized ribbon roughened by etching.These schemat- ics explain the mechanisms involved for processes that can cause features observed in Fig. 34. 84 After pre-oxidized, oxide layer \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\‘.s\\\\\ fibbon After acid etching, W After composite fabrication, matrix fibbon oxide layer During pull-out test. (b) Fig. 35 (continued) 85 shearing action during the pull-out tests. In addition, due to the relatively low ratio of the width to the depth of jagged areas at the interface, the fail- ure at the glass portion which fills the dimple on the ribbon surface can be expected before ribbon starts pulling-out. As a result, with this ribbon sur- face treatment, the pull-out stress can be obtained in a range between those of the composites reinforced with the as-received ribbon, and with the ribbon roughened by etching. Hence, this oxide layer helps in control- ling more precisely the density, size and depth of the dimples, as compared to those obtainable by etching of the as-received ribbon. Pull-out stresses of the composites containing ribbons with various sur- face treatments (for 1.7 mm-embedded ribbon length ) are given in Fig. 36. Etching of as-received ribbon for 60 seconds resulted in 100 MPa increase in pull-out stress, as compared to that for as-received or pre-oxidized rib- bons. Moreover, acid etching for 90 seconds resulted in matrix failure at the end of the embedded ribbon before ribbon starts to pull out due to high degree of interlocking at the interface. Slight increments in pull-out stresses (approximately 30 MPa) were obtained by etching the pre-oxidized ribbon for 60 seconds (as compared to as-received, and pre-oxidized rib- bons). In composites with as-received ribbon reinforcements, small load drops in stress-displacement curve representing that debonding front propagates in small jumps rather than continuously have been observed. Such a fea- ture was not found to be significant in this study since randomly spread out jagged edges of the interface cause delayed propagation of the debond- ing front with etched reinforcements. This results in continuous propaga- tion of the debonding front. 86 700 ._ 66—61:}110 5941122 A 600: 562i17.3 545:2” Q g 500: § :5 300: \ 2:... § only (60 secs.) (60 secs.) only after oxidized Ribbon Treatments Fig. 36 Pull-out stress of ribbons that have undergone various treat- ments for an embedded length of 1.7 mm. 87 4-3. Fracture Toughness Behavior 4-3-1. Effect of Fracture Toughness (ch) on Interfacial Bonding The calculated fracture toughness (K10) values of soda-lime glass matrix composites reinforced with Nichrome ribbons in various interfacial bond- ing conditions are given in Table 4. The interfacial bonding conditions uti- lized were as follows: 1) The specimens which were subjected to thermal cycling have weak interfacial bonding conditions. Moreover, it is impor- tant to note that as the number of thermal cycles increased, the interfacial bonding strength decreased. 2) The composite specimens reinforced with acid etched Nichrome ribbon showed strong interfacial bonding due to mechanical interlocking along the interface. This mechanical interlocking effect was lower in the composite with ribbons that were preoxidized prior to acid etching. Regardless of the interfacial bonding condition, an increase of up to 20% in fracture toughness was achieved for composites with 0.5 vol% reinforce- ment. Prestressing of the matrix due to the thermal expansion coefficient mismatch between the matrix and ribbon and crack shielding by the rib- bons may be attributed to the enhancement in fracture toughness (ch) of the present composite system. Since the thermal expansion coeflicient of the ribbon is larger than that of the glass matrix, the matrix is in compres- sion along the direction parallel to the length of the ribbon, thus providing a compressive zone which increases the level of applied stress required to open the crack (9,93,94). Crack shielding by the ribbon impedes the crack propagation and also changes crack front configuration (93,95,96). Fracto- 88 Table 4. Fracture toughness (ch) of soda-lime glass matrix reinforced Nichrome ribbon with various bonding conditions. Bonding Conditions vol% K1,,(MPa 5:) Matrix 0 0.98 i005 NT 0.8 1.16 $0.09 TC-3 0.5 1.22 10.11 TC-7 0.5 1.17 i016 AE-60 0.5 1.11 $0.04 FAB-60 0.5 1.17 i009 89 graphic evidence of the crack shielding by the ribbon is shown in Fig. 37. The interference fringes represent the path of the crack. Fig. 37 clearly shows that the ribbon impedes the motion of the crack as it crosses the rib- bon. As described by Green (95), the crack is pinned at both the edges of the ribbon, resulting in crack bowing. Following crack bowing, the second- ary crack initiates behind the ribbon and propagates fast to catch up with the primary crack. However, the interfacial debonding process which has been reported as a major mechanism of fracture toughness enhancement in fiber reinforced ceramic matrix composites was not observed during matrix cracking in the case of the present system. Catastrophic matrix failure, due to consider- ably low volume fraction of the ribbon in the composites, leads to negligible interfacial debonding during flexural loading. 4-3-2. Tensile Behavior of Pre-cracked Composite Typical stress-displacement curves (with nominal stress referring to load per unit ribbon cross-sectional area) obtained from the tensile tests on the pre-cracked composite specimens with various interfacial bonding con- ditions are given in Fig. 38. The results given in Fig. 38 are for composites reinforced with a ribbon of 12 mm embedded length (19, defined in Fig. 11). In this study, the composite specimens that have undergone 3 and 7 ther- mal cycles between room temperature and 650°C are referred to as TC-3, and TC-7 respectively. AE-60 and AE-9O are used to refer to 60 and 90 sec- ond acid etched ribbons in as-received conditions, and PAE refers to 60 sec- and acid etching of ribbons in preoxidized condition. NT is used for composites with ribbons in as-received condition. In the AE specimen (the 90 Crack Direction —————> Fig. 37 Fractographic observation of the crack shielding by the rib- bon. Interference fringes represent the path of the crack. Note: A: primary crack plane, B: ribbon, C: secondary crack plane. Crack direction refers to direction of crack movement. 91 A:AE BzNT CzTC—3 DzTC-7 NOMINAL STRESS(MPa) O ' ' I ' I ' I ' I I ' I ' F 1 O 2 4 6 8 10 12 14 16 DISPLACEMENT(rnm) Fig. 38 Typical stress-displacement curves obtained from the tensile loading of the pre-cracked composite specimens with various interfacial bonding conditions. 92 strongest bonding condition) the ribbon fractured at a considerably small displacement (curve ‘A’). As the bonding becomes weaker, the ribbon frac- tured at larger displacements as seen in curve ‘B’ for the non-treated com- posite specimen. The “gauge length” of the ribbon is confined to the debonded length along the embedded ribbon. Also, the debonded length decreases with an increase in interfacial bonding strength of the composite for a given applied stress. As a result, a general trend, in spite of some inconsistencies, indicating that the displacement at the point where the ribbon fractured decreased with an increase in the interfacial bonding strength can be noted in the data provided in Table 5. On the other hand, the ribbon pull-out from the matrix after complete debonding was observed in TC-3 and TC-7 specimens (curves ‘0’ and ‘D’, respectively). However, there was noticeable difference between these two cases. For TC-3 speci- mens, the interfacial bonding was strong enough to prevent complete deb- onding of the ribbon during initial stages of the test. As a result, the effective length of the bonded portion of the ribbon was large enough to sustain considerable plastic deformation, before complete debonding occurred. On the other hand, in the TC-7 specimens containing mostly inter-connected microcracks, the interfacial bonding was so weak that the debonding crack propagated more easily along the interface as compared to the TC-3 specimen. In addition to this, the low static frictional shear stress at the debonded interface (in TC-7) was not effective enough to hold the debonded portion of the ribbon. Therefore, complete debonding occurred at a much lower stress than that of the TC-3 specimen. Hence, the extent of plastic deformation was very low in the case of the TC-7 spec- imens. However, the displacement at a given stress level was considerably large when compared to that of the TC-3 specimen because of the differ- 93 Table 5. Effect of interfacial bonding on the displacement at the ribbon fracture. Bonding Condition TC'7 = Displacement at Ribbon Fracture 0.67 0.85 1.02 1.64 1.63 1.48 0.05 0.28 0.50 0.58 0.24 (mm) 94 ence in the debonded length (“gauge length”) of the ribbon between both cases. The heavy shaded portion observed in the case of TC-3 specimen (Fig. 38) indicates the “stick-slip” friction phenomenon. In this portion of the curve, each load drop was accompanied by an audible “click”. The mecha- nism of this phenomenon has been explained by Griffin et al. (50) and Deshmukh et al. (49). According to Griffin et al., the Poisson contraction and expansion of the fiber in combination with a difference between static and dynamic friction leads to stick-slip friction. In contrast, Deshmukh et al. pointed out that the release of excess strain energy stored in the free length of the fiber when the fiber reloads contributes to the above phenom- enon. As illustrated in curve ‘C’, the stress amplitude and wavelength of the stick-slip behavior decreases as the ribbon is progressively pulled out. This can be explained by the smoothening effect of the debonded interface during ribbon pull-out, since such a behavior predominantly depends on the asperities of the debonded surfaces at the interface. The stick-slip behavior was not observed in the TC-7 specimens due to the smooth fea- ture of the debonded interface, resulting in low frictional shear stress. 4-3-3. Toughness due to Crack Bridging (AG0) Fracture toughness due to crack bridging (AGO) in the composite under various bonding conditions was evaluated using Eq. 8 and the area under the stress-displacement curves. The results are given in Table 6. Enhancement in fracture toughness resulting from thermal cycling (curve ‘C’) of the composite [as compared to that of as-received ribbon (curve ‘B’), or acid-etched ribbon (curve ‘A’)] can be observed in Fig. 38. Plastic defor- 95 Table 6. Fracture toughness due to crack bridging (AGc) in various bond- ing conditions. AGc = Vf x area under the stress-displacement curve in Fig. 38. AG, (KJ/mz) Frictional sliding Plastic deformation Total NT 6.08 0 6.08 AE-60 2.87 0 2.87 TC-3 7.03 13.48 20.51 TC-7 3.25 7.67 10.92 96 mation and frictional sliding play important roles in the enhancement of fracture toughness due to thermal cycling (curve ‘C’). For as-received or acid-etched ribbons, the lower fracture toughness realized can be attrib- uted to the ribbon fracture prior to the onset of any extensive plastic defor- mation. It is evident from curves ‘A’ and ‘B’ that stronger the interfacial bonding the smaller is the area under the curve, due to the small displace- ment at ribbon fracture. Even though both the curves ‘C’ and ‘D’ are com- prised of plastic deformation and frictional sliding stages, the total area under curve ‘D’ is much smaller than that of the curve ‘0’, because of the low plastic deformation and low frictional shear stress involved in curve ‘D’. From these observations, it can be inferred that TC-3 specimen with 11mm-embedded length of the ribbon shows the maximum fracture tough- ness due to crack bridging (AGc). In addition, the effect of the embedded ribbon length on toughness of non-treated composite specimens is shown in Fig. 39. The ribbon fractured in the composite with over 12mm-embedded ribbon length, while ribbon pull-out occurred after plastic deformation of the debonded ribbon in the composite with less than 11mm-embedded rib- bon length. As a result, the maximum fracture toughness due to crack bridging (AGO) in the non-treated composite was achieved at 11mm-embed- ded ribbon length. Therefore, it can be defined as “critical embedded length (1a)” for the maximum fracture toughness due to crack bridging in a given bonding condition (and for the given ribbon thickness) since the rib- bon with this length effectively undergoes both plastic deformation and frictional sliding phenomena. The critical embedded length in various bonding conditions is shown in Table 7. Due to the wide variation in data, the results were presented as a range, instead of a single value. Neverthe- less, it can be concluded from the results of this study that the maximum 800— 1 =12mm l (D 700- 600- 500 400 300- 200- NOMINAL STRESS(MPa) 100- . I ' III 97 le=11mm III II . IIIII IIIIIIIIII I l‘ IMI "'II'III I'll” III 1'! IIIII II II I O I I I l O 2 4 I I I T'I‘I'I 6 8101214 DISPLACEMENT(mm) Fig. 39 Stress-displacement curves showing the effect of the embed- ded ribbon length on toughness (AGc) of NT composite speci- mens. 98 Table 7. Critical embedded length (18) for the maximum fracture tough- ness due to crack bridging in various bonding conditions. B°nfiing NT AE-60 PAE-60 TC-3 TC-7 Condltlons Critical Embedded 10~ 12 less less 10» 13 15~ 17 Length (mm) than 3 than 3 99 fracture toughness due to crack bridging can be achieved by using discon- tinuous metallic reinforcements with a length less than or equal to the crit- ical embedded length in the ceramic matrix composites. 5. Conclusions Metallic ribbons embedded in a ceramic matrix may debond from the matrix due to the combined effects of plastic behavior of the metallic ribbon and the distribution of shear stress along the embedded ribbon (shear lag model). Hence, the plastic behavior of the metallic ribbon was taken into account in analyses of ribbon pull-out in present study. The interfaces between Nichrome ribbon and soda lime glass matrix were modified by means of various numbers of thermal cycles, and various maximum tem- peratures used in thermal cycles. The interfacial bonding strength, com- plete debonding stress and frictional shear stress were measured by pull- out tests on thermally cycled specimens. Also, the interface of this compos- ite was modified by acid etching of as-received or oxidized ribbon surface. Finally, on the basis of above interface modifications, interfacial effects on mechanical properties of brittle matrix with ductile reinforcement were evaluated. The important observations are summarized below: 1) Plastic deformation of ductile ribbon provides significant contribution to the resultant interfacial shear stress (rt) at the interface in ductile ribbon reinforced brittle matrix composites. 2) Although the debonding process in this composite is catastrophic at the maximum required debond load, mechanical keying due to the roughness of the debonded surfaces may prevent a sudden load drop. 3) The applied tensile stresses required to initiate and complete debonding 100 101 (oi and Cf respectively) are independent of the embedded length in excess of a critical value. 4) Microcracks at the interface of 650-TC specimens are generated after 3 thermal cycles, they link together after 5 cycles, and make a completely connected crack after 7 cycles. A 500-TC treatment might not create enough thermal stresses to generate microcracking at the interface after 5 cycles. 6) The interfacial bonding strength decreases with increasing number of thermal cycles and drops significantly after 5 cycles of 650-TC. 7) Frictional shear stress on the debonded surfaces decreases with increas- ing number of thermal cycles due to smoothening effect which results from continual rubbing between the debonded surfaces during continued ther- mal cycling. 8) The complete debonding stress decreases with increasing number of thermal cycles because both contributors, initial debonding stress and fric- tional shear stress, decrease with increasing number of cycles. 9) The results of this study suggest that optimum interfacial bonding strength of soda-lime glass/Nichrome ribbon composite can be achieved by thermal cycling. 10) Acid etching of the ribbon gives rise to considerable improvement in pull-out stress of the composite. However, it is very difficult to control the 102 pull-out stress because ribbon is severely attacked by the acid even in very short time period. 11) A pre-oxidation treatment of the ribbon surface before etching can pro- tect the metallic ribbon from severe attack by acid so as to provide a better control of the ribbon pull-out behavior in the composite. 12) In composites with etched ribbons, randomly spread-out jagged inter- face causes delayed propagation of debonding front, and consequently deb- onding front propagates continuously rather than in small jumps, a feature exhibited by the composites reinforced with smooth ribbons. 13) Increases of fracture toughness (ch) of these composites up to 20% were achieved due to “prestressing of the matrix” and “crack shielding by the ribbon”. 14) Introduction of even a very low volume fraction of metallic ribbon pro- vides significant improvement in fracture toughness due to crack bridging (AGc) of the brittle glass matrix. 15) To obtain maximum toughness in this composite, it is essential to retain the optimum bonding strength. 16) The results of this study suggest that enhancement in toughness may be achieved by using discontinuous metallic ribbon reinforcement in the brittle. ceramic matrix (16 S lcr). References 1) DC. Phillips, J. Mater. Sci., 9 (1974) 1847 2) KM. Prewo and J .J . Brennen, J. Mater. Sci., 15 (1980) 463 3) J. Brennen and K. Prewo, J. Mater. Sci., 17 (1982) 2371 4) KM. Prewo and J .J . Brennen, J. Mater. Sci., 17 (1982) 1201 5) A.G. Evans and RM. McMeeking, Acta metall., 34 (1986) 2435 6) MD. Thouless and A.G. Evans, Acta metall., 36 (1988) 517 7) L.S. Sigl and HE Fischmeister, Acta metall., 36 (1988) 887 8) B. Budiansky, J .C. Amazigo and A.G. Evans, J. Mech. Phys. Solids, 36 (1988) 167 9) A.G.Evans and DB. Marshall, Acta metall., 37 (1989) 2567 10) A.G. Evans, Mater. Sci. Eng., A107 (1989) 227 11) L.S. Sigl and A.G. Evans, Mech. Mater., 8 (1989) 1 12) E. Bischofi', O. Sbaizero, M. Riihle, and A.G. Evans, J. Am. Ceram. Soc., 72 ( 1989) 741 13) A.G. Evans, J. Am. Ceram. Soc., 73 (1990) 187 14) S.V. Nair, J. Am. Ceram. Soc., 73 ( 1990) 2839 15) G.H. Campbell, B.J. Dalgleish, M. Riihle and A.G. Evans, J. Am. Ceram. Soc., 73 (1990) 521 16) GA. Cooper and A. Kelly, J. Mech. Phys. Solids, 15 (1967) 279 17 ) P. Hing and G.W. Groves, J. Mater. Sci.,7 (1972) 427 18) VD. Krstic, P.S. Nicholson and R.G. Hoagland, J. Am. Ceram. Soc., 64 (1981) 499 19) VD. Krstic, Philos. Mag. A48 (1983) 695 103 104 20) MS. Newkirk, A.W. Urqhart and HR. Zwicker, J. Mater. Res. 1 (1986) 81 21) L.S. Sigl, P.A. Mataga, B.J. Dalgeish, R.M. McMeeking and A.G. Evans, Acta metall., 36 (1988) 945 22) M.F. Ashby, F.J. Blunt and M. Bannister, Acta metall., 37 (1989) 1847 23) H.C. Cao, B.J. Dalgleish, H.E. Déve, C.K. Elliot, A.G. Evans, R. Mehra- bian and G.R. Odette, Acta metall., 37 (1989) 2969 24) B. Flinn, M. Riihle and A.G. Evans, Acta metall., 37 (1989) 2969 25) H.E. Déve, A.G. Evans, G.R. Odette, R. Mehrabian, M.L. Emiliani and R.J. Hecht., Acta metall. mater., 38 (1990) 1491 26) G. Baran, M. Degrange, C.Roques-Carmes and D. Wehbi, J. Mater. Sci., 25 (1990) 4211 27) KS. Ravichadran, Acta metall. mater., 40 (1992) 1009 28) FE. Heredia, M.Y. He, G. E. Lucas, A.G. Evans, H.E. Déve and D. Konitzer, Acta metall. mater., 41 (1993) 505 29) LB. Greszczuk, in " WW", ASTM STP 452 (Ameri- can Society for Testing and Materials,1969) p. 42. 30) P. Lawrence, J. Mater. Sci. 7 (1972) 1. 31) A. Takaku and R.G.C. Arridge, J. Phys. D: Appl. Phys. 6 (1973) 2038. 32) P. Bartos, J. Mater. Sci. 15 ( 1980) 3122. 33) C.W. Griffin, S.Y. Limaye, D.W. Richerson and D.K. Shetty, Ceram. Eng. Sci. Proc. 9 ( 1988) 671. 34) R.W. Goettler and KT. Faber, Ceram. Eng. Sci. Proc. 9 (1988) 861. 35) Chun-Hway Hsueh, Mater. Sci. and Eng. A123 (1990) 1. 36) V.M. Karbhari and D.J. Wilkins, Scrip. Metall. Mater. 24 (1990) 1197. 37) R.W. Rice, J .R. Spann, D. Lewis and W. Coblenz, Ceram. Eng. Sci. Proc., 5 (1984) 614 105 38) B. Bender, D. Shadwell, C Bulik, L. Invorvati and D. Lewisill, Am. Ceram. Soc. Bull., 65 (1986) 363 39) RA. Lowden and DP Stinton, Ceram. Eng. Sci. Proc., 9 (1988) 705 40) RN. Singh, Ceram. Eng. Sci. Proc., 10 (1989) 883 41) L.P.Zawada and R.C.Wetherhold, J. Mater. Sci., 26 (1991) 648 42) RH. Moore and SC. Kunz, Ceram. Eng. Sci. Proc., 8 (1987) 839 43) M.J. Klein, in "WW", 1: Interfaces in Metal Matrix Composites, edited by A.G. Metcalf, L.J. Broutman and RH. Krock (Academic, New York, 1974) 169 44) DH. Grande, J .F. Mandell and K.C.C. Hong, J. Mater. Sci. 23 ( 1988) 311 45) A. Kelly and W.R. Tyson, J. Mech. Phys. Solids, 13 (1965) 329 46) V. Laws, Composites, 13 (1982) 145 47) R.J. Gray, J. Mater. Sci., 19 (1984) 861 48) M.R. Piggott, Comp. Sci. Tech., 30 ( 1987) 295 49) U.V. Deshmukh and T.W. Coyle, Ceram. Eng. Sci. Proc., 9 (1988) 627 50) CW. Griffin, S.Y. Limaye, D.W. Richerson and D.K. Shetty, Ceram. Eng. Sci. Proc., 9 (1988) 671 51) Chun-Hway Hseuh, J. Mater. Sci. Letters, 7 (1988) 497 52) J. Kim, C. Baillie and Y. Mai, Scripta metall. mater., 25 (1991) 315 53) J. Kerans and T.A. Parthasarathy, J. Am. Ceram. Soc., 74 (1991) 1585 54) D.B. Marshall, J. Am. Ceram. Soc., 67 (1984) C-259 55) D.B. Marshall and W.C. Oliver, J. Am. Ceram. Soc.,70 (1987) 542 56) J .W. Laughner, N.J. Shaw, R.T. Bhatt and J .A. Dicarlo, presented at 10th Annual Conference on Composites and Advanced Ceramic Materi- als, American Ceramic Society, Coco Beach, FL, 1986 106 57) RD. Warren, T.J. Mackin and A.G. Evans, Acta metall. mater., 40 (1992) 1243 58) H.L. Cox, Brit. J. Appl. Phys. 3 (1952) 72 59) J. Bowling and G.W. Groves, J. Mater. Sci., 14 (1979) 431 60) J .D. Outwater and M.C. Murphy, in W .0 . O 0 [-g. :_ 01"¢'0 9“... ‘0 :f : once: ‘:.‘:cq The Society of the Plastics Industry, Inc. Washington DC. 1969, p. 11-c 61) D.K. Shetty, J. Am. Ceram. Soc., 71 (1988) C-107 62) J .D. Bright, D.K. Shetty, C.W. Griffin and S.Y. Limaye, J. Am. Ceram. Soc.,72 (1989) 1891 63) J .D. Bright, S. Danchaivijit and D.K. Shetty, J. Am. Ceram. Soc., 74 (1991) 115 64) T.P. Weihs and W.D. Nix, J. Am. Ceram. Soc., 74 (1991) 524 65) T.P. Weihs, O. Sbaizero, E.Y. Luh and W.D. Nix, J. Am. Ceram. Soc., 74 (1991) 535 66) C-H Hsueh, J .D. Bright and D.K. Shetty, J. Mater. Sci. Letters, 10 (1991) 135 67) R.J. Kerans and T.A. Parthasarathy, J. Am. Ceram. Soc., 74 (1991) 1585 68) D.B. Marshall, M.C. Shaw and W.L. Morris, Acta. metall.mater.,40 ( 1992) 443 69) D.B. Marshall, Acta. metall. mater. 40 (1992)427 70) W.C. Carter, E.P. Butler and E.R. Fuller, Scripta metall.mater. 25 ( 1991) 579 71) T.J. Mackin, P.D. Warren and A.G. Evans, Acta. metall. mater. 40 (1992) 1251 107 72) S. Small, R.P. Vozzola and LP. Zawada, J. Am. Ceram. Soc. 72 (1989) 1175 73) D.B. Marchall and B.N. Cox, Mech. Mech,Mater., 7 (1988)127 74) MD. Thouless, Acta. metall., 37 (1989) 2297 75) M. R’Mili, D. Rouby, G. Fantozzi and P. Lamicq, Comp. Sci. Tech., 37 (1990) 207 76) M. Bouquet, J .M. Birbis and J .M. Quenisset, Comp. Sci. Tech., 37 (1990) 223 77) W.F. Brown, Jr. and J .E. Slawley, W W ASTM STP 410 (American Society for Testing and Materials, 1966) 78) D.B. Marshall and A.G. Evans, J. Am. Ceram. Soc., 68 (1985) 225 79) M.F. Ashby, Acta metall. mater., 41 (1993) 1313 80) J .R. Rice, J. Appl. Mech., 35 (1968) 379 81) G.I. Barenblatt, Advances in Applied Mechanics, 7 Academic Press, 1962 82) J. Aveston, G.A. Cooper and A. Kelly, in W BMW, National Physical Laboratory, Tedding- ton, UK 1971 (IPC Science and Technology Press, Guildford, 1971) pp 15-24 83) D.B. Marshall, B.N. Cox and A.G. Evans, Acta. metall., 33 ( 1985) 2013 84) R.W. Davidge, Composites, 18 (1987) 92 85) B. Budiansky, J .W. Hutchinson and A.G. Evans, J. Mech. Phys. Solids, 34 (1986) 167 86) M. Sutcu and W.B. Hillig, Acta metall. 38 (1990) 2653 87) RN. Singh, J. Am. Ceram.‘Soc., 73 (1990) 2930 88) H.R. Schwietert and RS. Steif, J. Mech.Phys. Solids, 38 (1990) 325 108 89) G.W. Sherer and SM. Rekhson, J. Am. Ceram. Soc., 65 (1982) 399 90) KM. Prewo, J. Mater. Sci., 17 (1982) 3549 91) H. Mecking, in W WWW. Aachen. August 1979. edited by P- Haasen, V. Gerold and G. Kostorz (Pergamon Press,1979) 1573 92) J. Morton and G.W. Groves, J. Mater. Sci., 11 (1976) 617 93) R.W. Rice, Cer. Eng. Sci. Proc., 2 (1981) 661 94) R.L. Lehman and GA. Doughan, Comp. Sci. Tech., 37 (1990) 149 95) D.J. Green, P.S. Nicholson and J .D. Embury, J. Mater. Sci., 14 (1979)1413 96) R.U. Vaidya and K.N. Subramanian, J. Mater. Sci., 25 (1990) 3291