<%BANNER%>

Contact Fatigue Mechanisms as a Function of Crystal Aspect Ratio in Baria-Silicate Glass-Ceramics


PAGE 1

CONTACT FATIGUE MECHANISMS AS A FUNCTION OF CRYSTAL ASPECT RATIO IN BARIA-SILICATE GLASS-CERAMICS By KALLAYA SUPUTTAMONGKOL A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2003

PAGE 2

Copyright 2003 by Kallaya Suputtamongkol

PAGE 3

ACKNOWLEDGMENTS I would like to thank my advisors, Dr. K.J. Anusavice and Dr. J.J. Mecholsky Jr. I greatly appreciate the support and guidance they have provided for during my studies here at UF. They have also taught me valuable lessons of good research. I have had an excellent learning experience, which I can use to develop my career skills as a teacher. Special thanks go to Ben Lee who has always been there to help. Without Ben, I could not have gone home as soon as is now possible. Very special thanks go to my mother who has always encouraged me. I also would like to thank all my friends for their support and friendship. They have made my life here a lot easier and more pleasant. iii

PAGE 4

TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iii LIST OF TABLES.............................................................................................................vi LIST OF FIGURES..........................................................................................................vii ABSTRACT.......................................................................................................................ix CHAPTER 1 INTRODUCTION........................................................................................................1 2 LITERATURE REVIEW.............................................................................................4 Mechanisms of Fatigue Crack Growth in Ceramics.....................................................4 Mechanisms of Crack-Tip Shielding............................................................................6 Contact Fatigue.............................................................................................................9 Strength Degradation from Contact Damage.............................................................12 Environmental Effect..................................................................................................12 Fatigue Behavior of Dental Glass-Ceramics..............................................................15 Mechanism of Fatigue Damage..................................................................................18 Specific Aims..............................................................................................................20 3 MATERIALS AND METHODS...............................................................................21 Preparation of Baria-Silicate Glass-Ceramic Specimens...........................................21 Heat Treatment Schedules..........................................................................................21 Microstructural Analysis............................................................................................21 Surface Preparation.....................................................................................................22 Density........................................................................................................................22 Youngs Modulus and Poissons Ratio.......................................................................22 Test Methods..............................................................................................................23 Mechanical Fatigue Test......................................................................................23 Strength Degradation Test...................................................................................23 Fractographic Analysis...............................................................................................24 Statistical Analysis......................................................................................................25 iv

PAGE 5

4 RESULTS AND DISCUSSION.................................................................................26 Strength Degradation..................................................................................................29 Fractographic Analysis...............................................................................................30 Baria-Silicate Glass.............................................................................................30 Strength Degradation Associated with a Cone-Crack.........................................35 Baria-Silicate Glass-Ceramics.............................................................................45 Fatigue Failure Mechanisms.......................................................................................55 5 CONCLUSIONS........................................................................................................71 REFERENCES..................................................................................................................73 BIOGRAPHICAL SKETCH.............................................................................................78 v

PAGE 6

LIST OF TABLES Table page 2.1. Possible mechanisms of cyclic fatigue crack propagation in ceramics......................7 3.1. Heat treatment schedules..........................................................................................21 4.1. Physical properties of baria-silicate glass-ceramics.................................................26 4.2. Mean flexural strength and standard deviation for baria-silicate glass-ceramics....28 4.3. Crack sizes and geometric factors for glass calculated from equations 3.2 and equations 4.3-4.7......................................................................................................38 4.4. Crack sizes and geometric factors for glass specimens tested in deionized water.........................................................................................................................42 4.5. Geometric factors of glass specimens tested in deionized water with adjusted stress.........................................................................................................................43 4.6. Crack sizes and geometric factors of group AR 3 specimens tested in air..............47 4.7. Crack sizes and geometric factors of group AR 3 specimens tested in deionized water.........................................................................................................................47 4.8. Critical crack sizes of fractured group AR 8 specimens..........................................51 vi

PAGE 7

LIST OF FIGURES Figure page 2.1. Schematic illustration of the fatigue crack propagation stages..................................6 2.2. Hertzian contact loading...........................................................................................10 2.3. Cone crack parameters.............................................................................................13 2.4 Schematic illustration of the reaction of water and a strained Si-O-Si bond at the crack tip..............................................................................................................14 2.5. Schematic illustration of crack propagation in the presence of a chemical wedge.......................................................................................................................16 3.1. Schematic illustration of cyclic fatigue fixture........................................................24 3.2. Schematic representation of fracture surface features: a i and b i are original crack depth and width; and a cr and b cr are critical crack depth and width..............25 4.1. SEM images of baria-silicate glass-ceramic microstructures with aspect ratios of 3/1 (a) and 8/1 (b) after etching with 1% HF for 10 s (4000X)...........................27 4.2. Mean strength of baria-silicate glass-ceramics as a function of number of loading cycles...........................................................................................................29 4.3. Schematic illustrations of selected failure origins from cone cracks in glass and glass-ceramic specimens (Top view).......................................................................31 4.4. SEM images of a cone crack formed in deionized water after 10 4 loading cycles. Fracture typically originated from either point B........................................32 4.5. SEM images of a triangular crack in a glass specimen caused by cyclic loading in air for 10 5 cycles. Fracture initiated from points A-A.........................................33 4.6. SEM images of a semi-elliptical surface crack in a glass specimen caused by cyclic loading in deionized water for 10 5 cycles......................................................34 4.7. SEM images of a typical semi-elliptical crack in a glass specimen caused by polishing...................................................................................................................34 4.8. Schematic illustration of a cone crack induced by a Hertzian contact load.............36 vii

PAGE 8

4.9. Fracture origin associated with a Hertzian cone crack as suggested by Evans........36 4.10. Crack sizes as a function of cyclic loading for glass specimens: (a) Tested in air; (b) Tested in deionized water............................................................................39 4.11. SEM images of fracture surfaces of group AR 3 specimens subjected to cyclic loading from 10 3 to 10 5 cycles: (a) and (b) reveal semi-elliptical cracks and (c) and (d) show triangular cracks............................................................................48 4.12. Crack size as a function of loading cycles for AR 3 specimens...............................49 4.13. SEM images of fracture surfaces of group AR 3 specimens subjected to cyclic loading for 10 4 and 10 5 cycles in deionized water...................................................50 4.14. Crack size as a function of loading cycles for group AR 8 specimens tested in air and in deionized water........................................................................................52 4.15. SEM images of fracture surfaces of group AR8 specimens subjected to loading for 10 4 and 10 5 cycles..................................................................................53 4.16. SEM images of fracture surfaces of group AR 8 specimens subjected to loading for 10 4 and 10 5 cycles in deionized water...................................................54 4.17. Montage of SEM images (1000X) of fracture surfaces of group AR 3 specimen subjected to cyclic loading for 10 5 cycles in deionized water:................................57 4.18. SEM image of fracture surfaces of group AR3 specimens subjected to loading for 10 5 cycles in deionized water (2000X): (a) cone crack-related failure (higher magnification of Fig. 4.13a); (b) surface damage-related failure (higher magnification of Fig. 4.17).......................................................................................58 4.19. SEM images of surface contact damage in AR 3 specimens: (a) after 10 5 cycles in air; (b) after 10 5 cycles in deionized water...........................................................59 4.20. Images of surface contact damage in group AR 8 specimens (a) after 10 5 cycles in air; (b) after 10 5 cycles in deionized water...........................................................64 4.21. Montage of SEM images of fracture surfaces of group AR 8 specimens subjected to 10 5 cycles (1000X) in deionized water:...............................................65 4.22. SEM images of fracture surfaces of group AR8 specimens subjected to loading for 10 4 and 10 5 cycles in deionized water (3000X)..................................................66 4.23. SEM images of the cyclically loaded surfaces of baria-silicate glass-ceramics (a) AR 3 and; (b) AR 8 specimens tested in deionized water..................................67 4.24. SEM images of the cyclically loaded surfaces of baria-silicate glass-ceramics (a) AR 3 and; (b) AR 8 specimens tested in deionized water..................................68 viii

PAGE 9

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy CONTACT FATIGUE MECHANISMS AS A FUNCTION OF CRYSTAL ASPECT RATIO IN BARIA-SILICATE GLASS-CERAMICS By Kallaya Suputtamongkol May 2003 Chair: Kenneth J. Anusavice Cochair: John J. Mecholsky Jr. Major Department: Materials Science and Engineering Ceramic materials are potentially useful for dental applications because of their esthetic potential and biocompatibility. However, the existence of fatigue damage in ceramics raises considerable concern regarding its effect on the life prediction of dental prostheses. During normal mastication, dental restorations are subjected to repeated loading more than a thousand times per day and relatively high clinical failure rates for ceramic prostheses have been reported. To simulate the intraoral loads, Hertzian indentation loading was used in this study to characterize the fatigue failure mechanisms of ceramic materials using clinically relevant parameters. The baria-silicate system was chosen because of the nearly identical composition between the crystal and the glass matrix. Little or no residual stress is expected from the elastic modulus and thermal expansion mismatches between the two phases. Crystallites with different aspect ratios can also be produced by controlled heat treatment schedules. The objective of this study ix

PAGE 10

was to characterize the effect of crystal morphology on the fatigue mechanisms of baria-silicate glass-ceramics under clinically relevant conditions. The results show that the failure of materials with a low toughness such as baria-silicate glass (0.7 MPam 1/2 ) and glass-ceramic with an aspect ratio of 3/1 (1.3 MPam 1/2 ) initiated from a cone crack developed during cyclic loading for 10 3 to 10 5 cycles. The mean strength values of baria-silicate glass and glass-ceramic with an aspect ratio of 3/1 decreased significantly as a result of the presence of a cone crack. Failure of baria-silicate glass-ceramics with an aspect ratio of 8/1 (K c = 2.1 MPam 1/2 ) was initiated from surface flaws caused by either polishing or cyclic loading. The gradual decrease of fracture stress was observed in specimens with an aspect ratio of 8/1 after loading in air for 10 3 to 10 5 cycles. A reduction of approximately 50% in fracture stress levels was found for specimens with an aspect ratio of 8/1 after loading for 10 5 cycles in deionized water. The mechanisms for cyclic fatigue crack propagation in baria-silicate glass-ceramics are similar to those observed under quasi-static loading conditions. An intergranular fracture path was observed in glass-ceramics with an aspect ratio of 3/1. For an aspect ratio of 8/1, a transgranular fracture mode was dominant. x

PAGE 11

CHAPTER 1 INTRODUCTION Ceramic materials are potentially useful for a wide variety of medical and dental applications. They have several advantages over metals including their esthetic potential, chemical inertness, low thermal conductivity, low thermal diffusivity, excellent biocompatibility and their ability to be shaped using a variety of forming techniques. However, their major drawback is their low fracture resistance. For dental prostheses, relatively high clinical failure rates have been reported for several dental ceramics (1-3) and this failure potential limits their use for dental applications. During mastication, dental restorations are subjected to repeated loading, and crack initiation can occur after a significant number of stress cycles that can lead to fatigue failure. In an attempt to gain a better understanding of the failure mechanisms, previous investigators have focused on the characterization of fatigue damage in several dental ceramics (4-10). Hertzian indentation loading has been used in several studies to characterize the mechanical properties of ceramic materials (9-12). Contact indentation loading is a simple method for inducing surface damage and straightforward analytic relations can be derived for the critical loads in terms of basic material properties (i.e., elastic modulus, hardness, and fracture toughness) and contact radius (13). Clinically relevant parameters can be controlled by the use of a repeated indentation loading test to simulate occlusal intra-oral loads. However, the validity of the results for dental ceramics from the Hertzian indentation test may be questionable when the loading conditions for the test do not adequately reflect the loading characteristics that occur under situations in which 1

PAGE 12

2 clinical prostheses have failed. Based on previous studies, fracture initiation occurred at different sites depending upon the Hertzian loading conditions. The results of two clinical studies indicated that crack initiation of clinically failed fixed prostheses occurred either at the cement/ceramic interface or at the interface between the core and veneering ceramics (14, 15). In contrast, failure of specimens from a blunt indentation test occurs as a ring/cone crack or as a result of contact damage. The magnitude of load to failure values determined from the blunt indentation tests appear to be too high compared with normal occlusal forces. However, with a well-controlled testing protocol, valuable information about fatigue failure parameters and mechanisms can be obtained from this type of loading relative to potential dental applications. Cyclic fatigue tests on baria-silicate glass and glass-ceramics using a spherical indenter were selected to demonstrate the effect of crystal morphology on the mechanical response to contact fatigue damage. The baria-silicate glass-ceramic system was chosen for this study because (1) the system has a negligible difference in thermal expansion between the glass phase and crystal phase, thereby sustaining little residual stress upon cooling (16); (2) the baria-silicate glass phase and crystal phase have a nearly identical composition; (3) the baria-silicate system can be nucleated homogeneously without the addition of a nucleating agent; and (4) in higher crystal volume fractions, microstructures with different aspect ratios can be produced (17). The objectives of this study were to (1) test the hypothesis that after a specific number of cycles, the mean strength of baria-silicate glass-ceramics subjected to cyclic loading decreases because of critical flaw propagation; (2) test the hypothesis that the crack growth rate of baria-silicate glass-ceramics under cyclic loading and low stress

PAGE 13

3 conditions is strongly dependent on the number of stress cycles; (3) characterize the influence of crystal morphology on fracture behavior as a function of number of loading cycles; and (4) test the hypothesis that the crack growth rate of baria-silicate glass-ceramics subjected to cyclic fatigue can be accelerated by environmental effects.

PAGE 14

CHAPTER 2 LITERATURE REVIEW Recent studies (9-12) have demonstrated clearly that fatigue damage occurs under cyclic loading conditions in some commercial ceramics, and the strength of ceramics can be compromised by slow crack growth of a critical flaw during repeated loading. When the crack grows, the crack length increases over time until it reaches the critical value that can cause a catastrophic fracture. Therefore, stress values well below the levels that cause fast fracture can produce premature fracture of the ceramic structure. When the crack grows under cyclic loading conditions, the crack growth rate can be approximated using a simple power law equation that is dependent on the stress intensity ranges (18). To select ceramics for specific structural applications, one should be aware of the crack growth rates and cyclic fatigue behavior of ceramic materials during their clinical uses. Damage induced by cyclic loading in brittle solids can lead to much higher crack growth rates than those associated with environmental cracking under sustained loading conditions (6). As a consequence, the service lifetime of brittle solid structures can be reduced under repeated loading conditions. Mechanisms of Fatigue Crack Growth in Ceramics The crack growth behavior in ceramics is different than that in ductile materials. Dislocation slip, which is a prominent crack propagation mechanism in metals, generally does not occur in brittle solids at room temperature because of their covalent or ionic bonds (19-21). Subcritical crack growth in brittle solids under cyclic loading involves complex phenomena. In many brittle ceramics, there are no known differences between 4

PAGE 15

5 the mechanisms of static and cyclic crack growth at low temperatures. There are some general effects of cyclic loading on the growth of cracks, especially in ceramic materials containing long cracks, i.e., compared with their microstructures. The pronounced increase in crack growth rate caused by cyclic loading, compared with static test conditions, is primarily caused by the breakdown of the bridging zone in the wake of the crack tip. Residual stress produced by a thermal expansion mismatch between two phases and elastic anisotropy between the crystals can promote the formation of microcracks during the heating process and subsequent mechanical loading (18). The crack growth rate can be related to the stress intensity factor using a power law relationship: m)C(dNda (2.1) )]m[log(ClogdN dalog (2.1a) where da/dN is the crack growth rate per cycle, K is the stress intensity factor range (K max -K min ), and C and m are experimentally determined constants (18). The typical crack growth rate behavior of materials is represented by a sigmoidal curve on a logarithmic graph of da/dN versus K, and this curve can be divided into three distinct regions as shown in Fig. 2.1. The graph is linear in region II, which is consistent with equation 2.1. In previous studies, several crack growth mechanisms have been identified in ceramics that relate to intrinsic microstructural damage and extrinsic crack-tip shielding mechanisms (19-21). In metals, intrinsic mechanisms such as dislocation movement and crack-tip plasticity are dominant, and they involve blunting and resharpening of the

PAGE 16

6 crack. Conversely, extrinsic mechanisms result from formation of an inelastic zone surrounding the crack wake, or from physical contact with the crack surfaces. Region I Region II Region III m)C(dNda KIC Log K 10 -2 10-4 dNda 10-6 Figure 2.1. Schematic illustration of the fatigue crack propagation stages. Mechanisms of Crack-Tip Shielding Although shielding under monotonic loading results in a reduction in the local driving force, the effect under cyclic loading may be different because the magnitude of the principal driving force for fatigue crack progression is the difference between the applied maximum and minimum stress intensity factors, K. The difference in stress intensity factor can be increased or decreased by changing K max and K min While several mechanisms of cyclic fatigue crack growth in ceramics have been proposed, Ritchie et al. (20, 21) described the nature of ceramic fatigue according to two classes, intrinsic and

PAGE 17

7 extrinsic mechanisms. Summarized in Table 2.1 are the possible mechanisms for crack propagation under cyclic fatigue conditions in ceramics. Table 2.1. Possible mechanisms of cyclic fatigue crack propagation in ceramics. Mechanisms of cyclic fatigue crack growth in ceramics (1) Extrinsic mechanisms 1.1 Degradation of transformation toughening 1.2 Damage of bridging zone Friction and wear of unbroken ligaments or whisker/fiber reinforcements Crushing of asperities and interlocking zones 1.3 Fatigue of ductile reinforcing phases (2) Intrinsic mechanisms 2.1 Accumulated localized microplasticity/microcracking 2.2 Mode II and III crack propagation on unloading 2.3 Crack tip blunting/resharpening 2.4 Relaxation of residual stress Source (20) The shielding mechanisms involve crack deflection, zone shielding, contact shielding, and a combination of zone and contact shielding. These mechanisms are the principal means of increasing crack resistance in ceramics by reducing the local stress intensity range, K, which is a driving force for fatigue. Intrinsic mechanisms involve the creation of a fatigue-damaged microstructure ahead of the crack tip, while the extrinsic pathways involve a reduction of stress intensity via crack-tip shielding. Lathabi et al. (22) and Dauskardt (23) indicated that the frictional wear mechanism accounts for cyclic fatigue degradation in polycrystalline ceramics that exhibit grain bridging as a

PAGE 18

8 toughening mechanism. This model is based on the sliding wear degradation of frictional grain bridges, which reduces the toughening capacity from bridging under cyclic loading. The accumulation of debris from wear processes was observed at the frictional sliding sites during fatigue loading (24). It is suggested that this mechanism may be widespread in polycrystalline ceramics with a crack-bridging toughening mechanism. Brittle materials are known to fracture in monotonic compressive loading in two different patterns, extrinsic and intrinsic modes (25). In an extrinsic mode, cracks extend across planes of local maximum tension and propagate along a direction parallel to the maximum plane of compression. The intrinsic mode occurs by a microscopic process caused from an accumulation of pre-existing microcracks or weak interfaces and results in a shear fault when the critical loading conditions are met. Ewart and Suresh (26) confirmed the existence of crack growth in ceramics under cyclic compressive loading. They used single edge-notched polycrystalline alumina specimens subjected to a compressive cyclic load (26). Compressive stresses did not induce any crack growth, since only tensile stress can cause crack propagation. They found that the formation of particle debris within the crack was a characteristic feature of crack advance under cyclic compression. The particles were generated by fracture along the grain boundaries and their presence leads to crack closure effects. The authors reported that the crack growth rate increased in alumina specimens when they periodically cleaned the debris particles from between the crack faces. Therefore, their experimental results appear to be in conflict with previous data (19-21), which suggested that the formation of debris particles between the crack interfaces usually leads to a crack-tip opening condition or a wedging effect. However, the cleaning process may be responsible to an increase in crack growth

PAGE 19

9 rate by enhancing the stress corrosion at the crack tip. As a result, the increase in crack growth rate in this study can be accelerated by either a wedging effect or stress corrosion. The primary mechanism of cyclic fracture in alumina involves intergranular fracture, unlike quasi-static fracture in which a transgranular fracture mode is prevalent. Grain boundary microcracks nucleated during the compressive loading cycle within the process zone at the notch tip lead to a reduction in the elastic modulus of the material in that zone. The difference in the compliance of the material within the process zone and that of the surrounding area can induce residual tensile stress during unloading of the maximum compressive stress over a distance comparable to the size of the process zone. Ewart and Suresh (26) hypothesized that residual tensile stress contributes to the growth of fatigue cracks associated with grain boundary failure (18). Contact Fatigue For Hertzian contact of a sphere indenting a planar surface (18), one can calculate the contact radius (a) and the pressure (p) within the contact area following these assumptions: (1) the contact between the surfaces is nonconforming and frictionless; and (2) the dimensions of the two bodies in contact are significantly larger than the contact area. 3/1*4E3PRa (2.2) 2/1222)aR1(a23P p (2.3) 222121E1E1*E1 (2.4)

PAGE 20

10 where E 1 and E 2 are the elastic moduli, 1 and 2 are the Poissons ratio values of two bodies, and E* is the effective elastic modulus. The following general results are obtained from a Hertzian analysis (18, 27): (1) The maximum normal tensile stress occurs on the surface at the contact perimeter, and 0mp)21(21 where p 0 = P/a 2 (2) The maximum shear stress ( m ) exists beneath the surface along the contact axis at a depth of 0.48a, m = 0.48p 0 (3) These results imply that failure should initiate near the contact perimeter at the surface, while plastic yielding, crushing or microcracking is likely to occur beneath the surface. P 2R r Figure 2.2. Hertzian contact loading. Theoretically, when a hard spherical indenter is loaded normally on a flat, thick elastic specimen, it produces a classical Hertzian cone crack when the applied stress is greater than the critical value. The cone crack grows steadily with increasing load along the path that is normal to the tensile stresses. Roesler (28) established the following relation between the applied load P and cone base radius R:

PAGE 21

11 constant R P3/2 (2.4) The distinctive characteristic of a blunt indenter is an elastic contact zone during the time when the contact pressure increases monotonically with an expanding contact circle. The contact becomes inelastic only if the contact pressure exceeds some critical level for irreversible deformation prior to development of the cone-crack fracture (29). The microstructure of a material significantly affects the damage response. Elastic-to-plastic deformation occurs by microstructural coarsening of the crystal phase (9-12). A zone of residual stress can also be generated once permanent strain is induced during unloading (18). Residual stress can play a significant role in influencing the fatigue responses as well as failure processes. Fatigue can be considered either in terms of the accumulation of distributed damage or as the propagation of a single crack. Lawn et al (24) identified the damage model based on microcrack extension from the closed shear faults in the constraining compressive fields. These faults account for a quasi-plastic deformation at the site of loading that can lead to damage accumulation, material removal and wear, and strength degradation in ceramic materials. The onset of quasi-plastic deformation can be determined by the deviation from the nonlinear stress-strain curve (11, 12). Bonded-interface specimens were used to determine the severity of subsurface damage caused from a single indentation and contact fatigue. At the microstructural level, Lawn et al (27) identified the character of shear faults as (a) cracking at the weak interfaces between the glass phase and the second phase, (b) transgranular twinning, or (c) block slip in monolithic structures. The general feature of these faults is the discreteness, localized by the grain structure. The secondary microcracks are initiated at the ends of these faults.

PAGE 22

12 Strength Degradation from Contact Damage When blunt indentation produces the dominant flaws, the specimens will fail from the indentation sites. Examination of fractographic features from previous studies (27, 30) confirmed that failure of fine grain and medium grain ceramics originated from the cone base, and from within the subsurface damage zone in coarse-grain material. For brittle materials, the presence of a cone crack will cause a rapid reduction in critical stress for fast fracture of ceramic specimens. Strength degradation can be determined from the classical Irwin-Griffith equation by using crack size calculated from the radius of a ring crack, and a cone base angle. The critical stress intensity factor for the cone crack system is identified as (29) 3/2CcP (2.5) where c is the cone crack length, and is a dimensionless constant that depends only on Poissons ratio (29). The cone crack parameters are shown in Fig. 2.3. Although equation 2.4 has the same form as that for sharp indentation, the significance of the coefficient is different. For the sharp indentation, there is a residual stress field to accommodate the impression volume by expansion of the deformation zone against the constraining elastic matrix. Therefore, for sharp indentation is dependent on the nature of the deformation, elastic modulus, hardness of the material, and indenter half-angle. Environmental Effect The fracture resistance of glasses and crystalline ceramics can decrease over time under static loading and an active corrosive environment. The reduction in strength is associated with the slow crack growth of pre-existing flaws caused by a stress corrosion process. In silica, a chemical model for the interaction of the environment with

PAGE 23

13 mechanically strained bonds at the tip of a crack is proposed as a mechanism for slow crack propagation (31, 32). The process is described as consisting of the following events: (1) formation of a hydrogen bond between the water molecule and Si-O-Si bonds at the crack tip; (2) interaction of the lone pair of electrons from the oxygen atom in water (O w ) and the Si atom; (3) formation of two new bonds between O w and Si, and between the hydrogen and oxygen atoms from the silica molecule; and (4) rupture of the hydrogen bond with O w to yield surface Si-O-H groups on fracture surfaces. These Si-O-H groups will not reform without any external energy being supplied. P c C R0 0 Figure 2.3. Cone crack parameters. There are three regions of crack propagation in a graph of velocity (V) versus stress intensity factor (K 1 ). In region I, the crack propagation rate is sensitive to K 1 The sensitivity becomes moderate in region II and the crack growth is limited by the rate of water transport to the crack tip. The crack propagation rate is independent of the environment in region III. Subcritical crack growth can also occur in single crystal and

PAGE 24

14 polycrystalline ceramics (33-36). In alumina and zirconia, slow crack growth caused by water molecules has also been suggested (33-35). However, results from some studies suggest another approach to explain how the active environment affects the crack propagation rate (37-38). Si o Si Si oSi H o H Ho H Si o Si o HH Figure 2.4 Schematic illustration of the reaction of water and a strained Si-O-Si bond at the crack tip (32). Because of the atomic sharpness of the crack tip, an environmental molecule cannot penetrate to the crack tip and directly reacts with the chemical bond. Instead, a chemical wedge forms when chemisorbing molecules enter the opening of a crack (39). A crack propagating through a solid under the action of a chemical wedge is shown in Fig. 2.5. Besides static fatigue, failure that occurs by the simultaneous action of a cyclic stress and chemical attack is termed stress corrosion fatigue. In metals, mechanisms of environmental-assisted cracking are well documented (18, 40). Under cyclic loading, the embrittling environment can accelerate the propagation of a flaw to a certain critical size, and the crack propagation rate is enhanced as a result of exposure to a corrosive

PAGE 25

15 environment. The mechanisms of fatigue crack initiation and propagation involve accumulation of dislocations and electrochemical attack at the defective area. For ceramic materials, the detrimental effect of cyclic loading in a wet environment is well known (31-35). These results suggest that stress corrosion has a role in controlling overall crack propagation because of the faster growth rates in water than in air. The cyclic fatigue lifetime was shorter for alumina specimens tested in air compared with those tested in vacuum (41). This result indicates that corrosion fatigue can accelerate degradation associated with mechanical fatigue. In contrast, the results for Si 3 N 4 showed that the crack propagation mechanism under cyclic loading was independent of any stress corrosion mechanism (42). However, the results from another study indicated that the fatigue crack growth rate in Si 3 N 4 can be enhanced by a corrosive environment (43). Fatigue Behavior of Dental Glass-Ceramics Ceramics are routinely used for dental prostheses because of their superior esthetic qualities and biocompatibility. Because of the relatively low tensile and shear strength of dental porcelains, they are usually bonded to metal substructures made to fit prepared teeth to minimize the risk of fracturing the core or veneering ceramics. However, porcelain failures still occur and this represents a significant clinical problem for dental practices when ceramic restorations are needed (1, 2). Chipping or cracking of porcelain veneers has been reported to occur within 5 years after insertion (1, 2). Fracture is also a major drawback of ceramic systems that have been developed to improve the optical properties of ceramic prostheses. Relatively high clinical failure rates have been reported, ranging from 35% after three years to 64% after four years for tetrasilicic-fluormica-based glass-ceramic crowns cemented to molar teeth (3).

PAGE 26

16 CRACK PROPAGATION E D C B A Region A: Elastically distorted bonds Region B: Inelastic strain at the crack tip Region C: Ruptured bonds with secondary bonds Region D: Ruptured bonds without secondary bonds Region E: Bonds in equilibrium with the environment Figure 2.5. Schematic illustration of crack propagation in the presence of a chemical wedge (38). During normal mastication, dental prostheses are subjected to repeated loading more than a thousand times per day. At stress levels typically well below the ultimate stress of dental ceramics, unexpected fracture can occur after a significant number of cycles sufficient to cause fatigue failure of the restorations. Because of the brittle behavior of ceramic materials, they have little or no capacity to deform plastically under high loads. In the past, fatigue behavior was not expected to be an important factor in the

PAGE 27

17 control of failures. Recently, studies have demonstrated that dental ceramics can be gradually degraded under cyclic loading conditions (4-7). By using a sharp indenter, White (4) studied crack growth in feldspathic dental porcelain during repeated mechanical loading. He reported that fatigue behavior of this porcelain was consistent with the Paris model for cyclic mechanically induced crack growth (44). However, he suggested that this might not be an appropriate model for dental situations because of the high stress levels, which represents extreme conditions that occur rarely in vivo. Several studies used a blunt indentation technique to investigate the cyclic fatigue behavior of ceramics (5-7, 9). White et al. (5) used this technique in their study and they determined the critical stresses necessary to produce elastic-plastic deformation and crack initiation in feldspathic dental porcelain. They also confirmed the existence of fatigue damage by performing a cyclic loading test (5, 6). They concluded that cyclic mechanical fatigue caused irreversible damage to feldspathic porcelain. The severity of accumulated damage was quantified with respect to the gradual strength degradation after cyclic testing. It is well known that the presence of water will cause strength degradation of ceramic materials (8). White et al. (6) also investigated the effect of water on mechanical fatigue in feldspathic dental porcelain and reported results for ambient, wet, and dry environments. They reported that both static and cyclic mechanical fatigue significantly reduced specimen strength, but these two test conditions did not affect each other. Considerably greater damage and reduction in strength were noted for specimens tested in water.

PAGE 28

18 Besides feldspathic porcelain, other dental ceramic systems have also been investigated (7). Three ceramics, glass-infiltrated alumina (In-Ceram), a leucite-based glass-ceramic (Optimal Pressable Ceramic), and a leucite-based glass-ceramic (IPS Empress), were subjected to cyclic loading and then fractured in wet and dry environments. The decrease in fracture stress caused by cyclic loading was in agreement with the results of a previous study (6). This effect was more pronounced for specimens exposed to an aqueous environment. Mechanism of Fatigue Damage The existence of fatigue damage in ceramics raises considerable concern regarding the life prediction of brittle materials. It has been shown that subcritical crack growth occurs under cyclic loading conditions with stress levels well below the critical value for catastrophic fracture. The nature of crack propagation and crack growth resistance of these brittle materials has been widely investigated. The results of these studies suggest that several factors and mechanisms are involved in the fatigue behavior of different types of ceramic materials (45, 46). For dental ceramics, Peterson et al. used bonded-interface specimens to evaluate the role of microstructure and related failure mechanisms of specimens subjected to repeated loading (9, 10). A series of micaceous glass-ceramics were prepared by controlled heat treatment with mica crystal diameters between 1 and 10 m and aspect ratios between 3/1 and 9/1. The polished surfaces of two ceramic block specimens were bonded together with cyanoacrylate adhesive. After cyclic loading, two damage modes were observed, tensile-induced cone cracking in a fine-grain material and quasi-plastic deformation in a coarser grain glass-ceramic. They concluded that microstructure is a

PAGE 29

19 controlling factor in determining the nature and degree of damage accumulation in dental ceramics. They also used other dental glass-ceramics in their study, such as Vita In-Ceram, a glass-infiltrated alumina, Vita Mark II, a modified feldspathic porcelain, and yttria-stabilized tetragonal zirconia. They found that a quasi-plastic response was observed in zirconia and glass-infiltrated alumina. The most brittle response (cone crack formation) was observed in modified feldspathic porcelain. The bonded-interface technique has been used in earlier studies (11, 12) in which the damage beneath a Hertzian contact field was examined in commercial mica-containing glass-ceramics compared with the glass specimens. For the glass samples, fatigue failure was caused by slow growth of cone cracks. On the other hand, the fatigue process in glass-ceramics was enhanced by formation of microcracks developed from shear stress along the weak interfaces between the mica platelets and the glass matrix. The effect of environment on the fracture behavior of dental ceramics has also been studied (47-50). Specimens made from aluminous and feldspathic porcelain were tested under three-point loading (49). The authors found that specimens that were tested under dry conditions were approximately 27% stronger than those that were tested in distilled water. Previous studies showed that dental ceramics could be weakened by fatigue damage (4-7). The investigators attempted, with little success, to identify the mechanisms responsible for the degradation of the materials during cyclic fatigue. They also used a tungsten carbide indenter to produce contact damage within the specimens. Tungsten carbide has a remarkably higher elastic modulus than most other materials that are used to restore missing tooth structure. Thus, the use of a tungsten carbide indenter

PAGE 30

20 can be considered as an extreme condition and the experimental results might not be representative of the oral environment. Several previous investigators analyzed the fracture failure of dental ceramics in a brittle mode because they thought that ceramic materials could not deform plastically. Therefore, only a limited number of investigators have analyzed the fatigue damage caused from repeated loading conditions. The lack of information in this area is obvious. In addition, all previous data has been obtained on materials that have thermal expansion anisotropy, which leads to development of a residual stress upon cooling. The residual stress may affect the damage modes observed. Thus, a study that can minimize extraneous effects from microstructural property is preferred. Specific Aims The specific aims of this study were to (1) test the hypothesis that, at a specific number of cycles, the mean flexural strength of baria-silicate glass-ceramics subjected to cyclic loading is rapidly decreased because of critical flaw propagation; (2) identify the controlling mechanism of crack propagation in baria-silicate glass-ceramics based on two crystal aspect ratios; (3) test the hypothesis that the crack growth rate of baria-silicate glass-ceramics under cyclic loading and low stress conditions is strongly dependent on the number of stress cycles; (4) characterize the influence of crystal morphology on fatigue failure as a function of number of loading cycles; and (5) test the hypothesis that the crack growth rate of baria-silicate glass-ceramics subjected to cyclic fatigue can be accelerated by environmental effects.

PAGE 31

CHAPTER 3 MATERIALS AND METHODS Preparation of Baria-Silicate Glass-Ceramic Specimens Baria-silicate glass plates containing 39.5 wt% BaO and 60.5 wt% SiO 2 were obtained from Corning Inc. (Corning, NY). Bars, approximately 25 mm long, 5.1 mm wide, and 2.5 mm thick, were cut from these plates using a diamond cutting saw (Isomet, Buehler IL). Heat Treatment Schedules The glass bars were heat-treated to produce glass-ceramic specimens with desired microstructures. One hundred and forty bars were nucleated and crystallized to produce two different microstructures. The treatment times and temperatures are listed in Table 3.1. Seventy glass specimens that were not heat treated were used as control specimens. Table 3.1. Heat treatment schedules. Nucleation Crystallization Temp (C) Time (h) Temp (C) Time (h) Aspect ratio 700 0.75 825 1.5 3.6 700 1.0 825/1050 24/24 8.1 Microstructural Analysis The percent crystallization of baria-silicate glass-ceramics was determined using the point-counting technique. Specimens were etched in 1% HF for 10 s, rinsed in deionized water, dried and coated with Au-Pd for scanning electron microscopy. Sixteen 21

PAGE 32

22 images for each aspect ratio group were recorded [4 specimens x 2 cuts (90 and 45) x 2 surface areas]. The aspect ratios were determined from the major and minor dimension of discernible crystals in SEM micrographs. All micrographs were produced at 4000X. Surface Preparation Bar-shaped specimens, 25 mm long, 5 mm wide, and 2 mm thick, were prepared from baria-silicate glass-ceramics. The glass and glass-ceramics bars were ground and polished through 600-grit silicon carbide abrasive using a hand lapper (South Bay Technologies, San Clemente, CA). The surfaces to be subjected to cyclic fatigue loading were polished through 1200-grit silicon carbide and 1 m alumina paste. The specimen edges were rounded using 1200-grit silicon carbide to reduce the risk of premature failure at stress concentration sites. All specimens were annealed at 600 C for 2 h after polishing, and ultrasonically cleaned in ethanol for 5 min before testing. Density The density of each material was determined from the mass to volume ratio. Each specimen was dried in furnace and its dry weight was measured on a precision balance (Mettler H31, Mettler-Toledo Inc., NJ). Three volumetric readings were made for each specimen to determine the average volume using a pycnometer (Model 1330, Micrometrics, Norcross, GA). The density of each specimen was then calculated. Youngs Modulus and Poissons Ratio Groups of four specimens each were prepared for each aspect ratio and used to determine elastic modulus and Poissons ratio values using an ultrasonic device (Nuson Inc., Boalsburg, PA). The transducers were attached to the specimens to generate the

PAGE 33

23 longitudinal and shear waves. Elastic modulus and Poissons ratio were calculated from the shear and longitudinal velocity values. Test Methods Mechanical Fatigue Test A cyclic load was applied to baria-silicate glass-ceramic specimens using a servo-hydraulic testing machine (Model 1350, Instron Corp., Canton, MA). A schematic illustration of the cyclic fatigue fixture is shown in Fig. 3.1. In an attempt to simulate test conditions that approximate oral conditions, a low loading frequency (3 Hz) and a low load level ( 200 N) were used. A blunt Type 302 stainless steel indenter was used to apply the load on the specimen surface. The diameter of the indenter used in this study was 4.76 mm, which is approximately equal to the cuspal radii of molar and premolar teeth. Ten bar specimens were subjected to one of the following numbers of loading cycles: 0, 10 3 10 4 and 10 5 cycles. After cyclic loading, specimens were observed for surface damage using an optical microscope, and then they were subjected to four-point bending and observed for subsurface damage. Strength Degradation Test A four-point bending test was used to quantify the severity of the strength reduction caused by the repeated loading test. Glass-ceramic specimens were fractured using a universal testing machine (Instron Model 1125) at a crosshead speed of 2.5 mm/min. A four-point flexure apparatus with an upper span of 6.67 mm and a lower span of 20 mm was used. The load to failure, P, was used to calculate the fracture stress, f using the following equation: 2f b t3Pa (3.1)

PAGE 34

24 where P is the failure load, a is the horizontal distance between support and loading points, b is the bar width, and t is the bar thickness. P: 0-200 N Loading cylinder Steel ball Specimen bar Steel support Figure 3.1. Schematic illustration of cyclic fatigue fixture. The environments used for the cyclic fatigue test were ambient room temperature (25C) and deionized water at 37C. Fractographic Analysis The critical flaw size was measured using an optical microscope. The fracture toughness of each material was determined from specimens in the control group (0 cycle) using the Griffith-Irwin equation (51): (3.2) 1/2c(c)Y where Y is a geometric constant, is the calculated fracture stress, and c is the measured crack size within the fracture surfaces. The crack depth was calculated as the square root of the depth times the half width. A schematic illustration of fractographic features is

PAGE 35

25 shown in Fig. 3.2 (52). Surface and subsurface damage of specimens subjected to cyclic loading were identified using an optical microscope and SEM images. Statistical Analysis An analysis of variance was performed to determine if a statistically significant difference existed between mean values of the independent variables. The null hypothesis (H 0 ) was that there was no statistically significant difference between mean values. The alternative hypothesis (H A ) was that not all means were equal. The significance level () was set at 0.05. P-values were calculated from the data. If the p-value was lower than the significance level ( = 0.05), the null hypothesis was rejected, and a multiple comparison analysis was performed to determine the order of the statistical subsets. Hackle region Mist region Mirror region Critical crack c = (ab)1/2 Figure 3.2. Schematic representation of fracture surface features: a i and b i are original crack depth and width; and a cr and b cr are critical crack depth and width.

PAGE 36

CHAPTER 4 RESULTS AND DISCUSSION Physical properties of baria-silicate glass and glass-ceramics are summarized in Table 4.1. Youngs modulus increased from 64.8 GPa for the glass to 80.5 GPa for the glass-ceramic with a high aspect ratio. These values are in good agreement with those reported by Freiman et al. (17). Table 4.1. Physical properties of baria-silicate glass-ceramics. Materials Aspect Ratio Density (g/cm 3 ) Youngs Modulus (GPa) Poissons Ratio Crystal Volume Fraction Glass 3.99 0.01 64.8 0.29 0% Glass-ceramic (AR 3) 3.6 3.93 0.02 75.3 0.28 75% Glass-ceramic (AR 8) 8.1 4.02 0.01 80.5 0.27 73% The highly crystalline (> 70 vol%) glass-ceramics were produced with two aspect ratios (AR) using heat treatments specified in Table 3.1. For glass-ceramic materials, the crystal diameter ranged from 1.5 to 11 m for the AR3 group, and from 0.5 to 15 m for the AR 8 group. The microstructures of the two glass-ceramics are shown in Fig. 4.1. The crystals in AR 3 group specimens appear spherulitic with an aspect ratio of 3.6:1. Crystals in the AR 8 group specimens are elongated with an approximate length to width aspect ratio of 8:1. 26

PAGE 37

27 (a) AR 3 (b) AR 8 Figure 4.1. SEM images of baria-silicate glass-ceramic microstructures with aspect ratios of 3/1 (a) and 8/1 (b) after etching with 1% HF for 10 s (4000X).

PAGE 38

28 Table 4.2. Mean flexural strength and standard deviation for baria-silicate glass-ceramics. Number of cycles Environment Glass (MPa) AR 3 (MPa) AR 8 (MPa) 0 Air 96 14 (A) 150 20 (A') 230 14 (A") 10 3 Air 61 20 (B) 138 20 (A') 225 12 (A") 10 4 Air 63 19 (B) 108 33 (B') 208 20 (B") 10 5 Air 51 14 (B) 80 18 (C') 209 17 (B") 0 Deionized water 98 17 (a) 150 20 (a') 185 7 (a") 10 3 Deionized water 53 13 (b) 118 24 (b') 201 19 (a") 10 4 Deionized water 45 11 (b) 100 14 (c') 171 24 (b") 10 5 Deionized water 50 9 (b) 82 15 (d') 110 15 (c") ( ) The mean values in a column with the same letter are not significantly different

PAGE 39

29 0501001502002503000100010000100000Loading cyclesFlexure strength (MPa) Figure 4.2. Mean strength of baria-silicate glass-ceramics as a function of number of loading cycles ( Tested in air; Tested in water). AR 8 AR 8 W AR 3 AR 3 W Glass Glass W Strength Degradation The mean flexural strength and standard deviation for baria-silicate glass and glass-ceramic specimens that were cyclically loaded at ambient room temperature and in deionized water are summarized in Table 4.2. The strength variation as a function of number of loading cycles is shown in Fig. 4.2. Based on Duncans multiple comparison tests, the mean strength of the group with an aspect ratio of 8/1 (AR 8) was significantly greater than that of each of the other groups at each number of cycles in air (p 0.05). The mean flexural strength of group AR 3 was also significantly greater than the mean of the glass group at each number of cycles in air (p 0.05). The mean flexural strength of glass specimens decreased significantly after 10 3 cycles (p 0.05). For glass-ceramic specimens with the lower

PAGE 40

30 aspect ratio (AR 3), the mean strength decreased after loading for 10 4 cycles (p 0.05). A reduction in mean strength of group AR 8 specimens was observed after 10 4 and 10 5 cycles. The decrease in strength of baria-silicate glass and AR3 specimens that were cyclically loaded in deionized water shows a similar trend compared with those exhibited by the specimens tested at room temperature, but the decrease in strength occurred at fewer cycles. For the glass and AR 3 groups, the strength decreased after loading for 10 3 cycles. The decrease in mean strength of specimens with an aspect ratio 8/1 was observed at 10 3 cycles. The mean strength of group AR 8 specimens tested in air did not decrease as much as the mean strength of glass and group AR 3 specimens. However, in deionized water, the decrease in strength of group AR 8 was significant. After 10 5 cycles, a decrease of approximately 50% in mean strength was observed for AR 8 group specimens. Fractographic Analysis Baria-Silicate Glass Fractographic analysis can be used as a characterization tool to detect the type of strength-limiting cracks formed by cyclic loading. Based on fracture surface analysis of cyclically loaded baria-silicate glass specimens, four crack configurations were observed and a schematic illustration of two of these failure origins is shown in Fig. 4.3. The first type (I) was a concentric ring and cone crack type (Fig. 4.4), which initiated from the base of the cone crack and propagated approximately perpendicular to the principal direction of tensile stress (point B in Fig. 4.3). The second type (II) was a triangular-shaped crack that formed during four-point flexure at the base of the cone crack (point A-A in Fig. 4.3) and subsequently propagated along the plane of a cone that developed

PAGE 41

31 during cyclic loading (Fig. 4.5). The third type (III) was a semi-elliptical surface crack caused by cyclic loading damage at the site of loading (Fig. 4.6). The fourth crack configuration (IV) was a semi-elliptical crack caused by polishing (Fig. 4.7). The first two types were observed in the glass and AR 3 specimens subjected to cyclic loads ranging from 10 3 to 10 5 cycles. The third crack type was observed in group AR 3, AR 8 and glass specimens tested in deionized water for 10 4 and 10 5 cycles. The fourth crack type was observed in specimens in the control group (0 cycle) and in specimens in which the damage from cyclic loading was not as severe as the surface finishes. a a a (1) E E D A' A A a B B (2) (2) Ring crack C C Cone crack (1) Triangular crack originated from point A-A and propagated to point A' on a AABED plane to form a critical crack leading to fast failure (2) Failure started at either of the two point B sites Figure 4.3. Schematic illustrations of selected failure origins from cone cracks in glass and glass-ceramic specimens (Top view).

PAGE 42

32 (a) 25X (b)80X Figure 4.4. SEM images of a cone crack formed in deionized water after 10 4 loading cycles. Fracture typically originated from either point B.

PAGE 43

33 (a) Loading and fracture surface images (100X) Fracture surface Fracture surface Triangular crack A A A (b) Fracture surface images (30X and 100X) Figure 4.5. SEM images of a triangular crack in a glass specimen caused by cyclic loading in air for 10 5 cycles. Fracture initiated from points A-A.

PAGE 44

34 Critical flaw (a) 25X (b) 100X Figure 4.6. SEM images of a semi-elliptical surface crack in a glass specimen caused by cyclic loading in deionized water for 10 5 cycles. The dashed line in (b) outlines the region of the critical crack. Critical flaw Fracture surface (a) 75X Critical flaw (b) 200X Figure 4.7. SEM images of a typical semi-elliptical crack in a glass specimen caused by polishing.

PAGE 45

35 Strength Degradation Associated with a Cone-Crack Few studies have been reported on the analysis of strength degradation caused by a Hertzian indentation crack (53,54). For well-developed cone cracks (Fig. 4.8), dimensional analysis (29) indicates that: c3/2K R P (4.1) where P is the indentation load, R is the radius of the cone base, and (, ) is a dimensionless constant dependent on Poissons ratio () and the cone-crack angle (). K c is the toughness of the material in term of the critical stress intensity factor. Lawn et al. described the strength degradation ( f ) from cyclic loading related to the cone base radius (Fig. 4.8) as (53): 2/1fCR (4.2) 2f2C)()K(R)( (4.3) where () is a dimensionless constant dependent on Evans (54) assessed the effect of conical cracks on the strength of ceramic components (Fig. 4.9). He analyzed the stress necessary to extend a Hertzian crack, which should be inversely proportional to the square root of the crack depth, L. )L(ZY'K2/1fC (4.4) )KL(Y'ZC2/1f (4.5) where L = G sin Y' = 1/2 and Z is a flaw morphology parameter.

PAGE 46

36 P 2R 2a Figure 4.8. Schematic illustration of a cone crack induced by a Hertzian contact load. a a A L G Figure 4.9. Fracture origin associated with a Hertzian cone crack as suggested by Evans (54). Failure initiates at point A.

PAGE 47

37 An analysis of the stress intensity factor for a triangular crack inclined at different angles to the tensile axis was reported by Murakami (55). The crack area projected in the direction of the maximum tensile stress was used to calculate the stress intensity factor along the crack front using the following equation: K (4.6) 2/12/1P0])A([Cmax 2/12/1P2/10])A[(Kmax C (4.7) where A p is the crack area projected in a plane perpendicular to the direction of the tensile stress and C 0 is a constant obtained from the above equation. K max is the critical stress intensity factor in the polar coordinate system (r, ). Based on fracture surface analysis, the procedures used to measure the crack sizes and calculate geometric constant values were described (equations 4.3 4.7). Summarized in Table 4.3 are crack sizes and geometric constants calculated from the known fracture toughness of baria-silicate glass (0.7 MPam 1/2 ), and equations 4.3, 4.5 and 4.7 for the triangular crack (the second crack type). Critical crack sizes were also calculated using the Griffith-Irwin equation (equation 3.2) and Y values are included in Table 4.3 (51). The crack size as a function of loading cycles of glass specimens is shown in Fig. 4.10. Few studies have analyzed the potential methods to predict strength degradation caused by cone cracks (53-56). Based on the cone crack configuration, failure in flexure is expected to initiate from one of the two diametrically opposite positions on the base rim of the cone (point C-C in Fig. 4.3) in the symmetric plane containing the axes of indentation loading. The failure analysis for this condition was developed in terms of

PAGE 48

38 fracture mechanics principles, which assume that the Hertzian cone crack parameter, R(), is the critical flaw size (53). The approximation of strength degradation by the cone crack as described in another study was also considered (54). Although the author (54) agreed that the fracture should originate from a position at the base exposed to the maximum stress intensity (point A in Fig. 4.9), the decreased strength values were actually determined by using the depth of the cone as the critical crack size. Based on the fracture surface analysis in the present study, we observed two crack configurations that could not be clearly characterized by the methods described previously (53-54). For example, the observed triangular crack that was likely formed during four-point flexure. From the fracture surface markings, we observed two symmetric points on the cone base as failure origins. Cracks propagated from these two points to join each other and form a critical crack for catastrophic failure. Thus, failure of specimens in this study did not start from the site (point C-C in Fig. 4.3) described in previous studies (53-54) and resulted in formation of a different crack configuration. Table 4.3. Crack sizes and geometric factors for glass calculated from equations 3.2 and equations 4.3-4.7. No. of cycles R (m) (Eq. 4.3) L (m) (Eq. 4.5) A p (m) (Eq. 4.7) (ab) 1/2 (m) (Eq. 3.2) Y'/Z (Eq. 4.5) C 0 1/2 (Eq. 4.7) Y (Eq. 3.2) 10 3 (9) 54 12 120 5 109 10 113 7 1.18 0.16 1.24 0.10 1.22 0.12 10 4 (8) 56 11 130 16 118 15 128 8 1.15 0.10 1.17 0.07 1.17 0.11 10 5 (9) 64 36 149 17 153 13 144 22 1.13 0.27 1.11 0.28 1.15 0.26 The mean critical crack size for the control group (0 cycle) is 35 14 m. ( ) Number of specimens in each group

PAGE 49

39 (a) Air 0501001502002500100010000100000Loading cyclesCrack size (m) R L Ap (ab)1/2 (b) Deionized water 0501001502002500100010000100000Loading cyclesCrack size (m) R L Ap (ab)1/2 R: Crack size for equation 4.3; L: Crack size for equation 4.5 A p : Crack size for equation 4.7; (ab) 1/2 : Crack size for equation 3.2 Figure 4.10. Crack sizes as a function of cyclic loading for glass specimens: (a) Tested in air; (b) Tested in deionized water. The legends in the graphs refer to the parameters used in Eqs. 3.2, 4.3, 4.5, 4.7 and in Table 4.3.

PAGE 50

40 As a result, it is unrealistic to use the methods described in those studies to characterize the triangular crack. Therefore, comparisons between crack size and the geometric constant values determined from each method have been made to determine the optimal technique for describing this crack system. From Table 4.3, the mean crack sizes calculated from equations 3.2, and 4.7 [(ab) 1/2 A p ] were comparable and not significantly different (p > 0.05). Y and C 0 1/2 values were also comparable. The crack size calculated from Equation 4.3 was significantly smaller than that calculated from the other equations. Lawn et al. (53) estimated that () values range from 0.25 to 0.30 for soda-lime silica glass. The () values of 0.11 for soda-lime glass and 0.34 for alumina were obtained from another study (56). Because of different () values reported for different materials, they suggested that () is a material constant (56). In our study, the () value was similar to that reported by Lawn, but with a larger standard deviation (SD). The reason for the large SD is most likely associated with the failure of glass specimens that does not initiate from one of the two diametrically opposite positions (either point C in Fig. 4.3) on the base rim of the cone coincident with the symmetry plane containing the axes of initial contact loading (53). The second possible reason is that the curvature of the cone base must be considered when the cone crack is large. The Y'/Z values calculated from equation 4.5 are comparable to Y and C 0 1/2 values. Evans reported that the Z value ranged from 1.3 to 1.5 for glass, alumina and silicon carbide (54). He also suggested that the values obtained in his study were similar to that for a surface crack of similar depth with a length to depth ratio of 3/1. For a

PAGE 51

41 length to depth ratio of 3/1, the value of flaw parameter (Q) is 1.75 (with a/b = 0.3). The Q value can be related to the critical stress intensity factor (40): QaYc (4.8) The Z value in equation 4.5 is reported to be a dimensionless constant that depends on the configuration of the crack (54), i.e., a correction factor for a semi-elliptical crack. However, the approach using this Z value and the depth of the critical crack seems to be inappropriate because only one dimension of a cone crack is considered. For critical crack size determination, width and depth are usually included to obtain an accurate crack size value. The location of the point at which fracture initiates is also critical in determining the crack size and a geometric constant value and should be identified during failure analysis processes. Summarized in Table 4.4 are the mean crack sizes and standard deviations determined from equation 3.2 and equations 4.3-4.7 of baria-silicate glass specimens that were cyclically loaded in deionized water. We observed the same crack patterns in the groups subjected to 10 3 and 10 4 loading cycles. However, contact damage within the surfaces of four specimens in the 10 4 cycle group and eight specimens in the 10 5 cycle group was clearly evident and different from that observed in specimens tested in air. Failure of these specimens tested in deionized water initiated from a semi-elliptical crack (type III) caused by contact surface damage (Fig.4.6). The mean critical crack sizes of these groups (10 4 cycle and 10 5 cycle group) were 194 55 and 135 32 m, respectively. From Table 4.4, the values of Y, Y'/Z, and C 0 1/2 tend to decrease gradually with an increase in the number of cycles from 10 4 cycles to 10 5 cycles. We hypothesize that the

PAGE 52

42 depth of the crack tip is responsible for reduction in these three constant values, i.e., there is a decrease in tensile stress, as the cracks grow deeper into the structure. The cone crack in specimens tested in deionized water propagated to a greater depth beneath the contact surface than that in specimens tested in air because of the stress corrosion effect. Therefore, critical crack sizes of specimens tested in deionized water were larger and resulted in a decrease in constant values. We recalculated the stress values according to the depth of the critical cracks, cal, using the following equation: ) b 2x1(fcal (4.9) where f is the stress at fracture, b is the thickness of a specimen, and x is the depth of the critical crack. Table 4.4. Crack sizes and geometric factors for glass specimens tested in deionized water. No. of cycles R (m) (Eq. 4.3) L (m) (Eq. 4.5) A p (m) (Eq. 4.7) (ab) 1/2 (m) (Eq. 3.2) Y/Z (Eq.4.5) C 0 1/2 (Eq.4.7) Y (Eq.3.2) 10 3 (9) 54 17 138 39 105 9 104 10 1.12 0.17 1.27 0.21 1.27 0.21 10 4 (4) 71 29 163 8 149 11 147 14 1.14 0.25 1.21 31 1.21 0.31 10 5 (2) 52 12 169 24 166 14 168 0.6 0.98 0.18 0.98 0.11 0.98 0.11 The mean critical crack size for the control group (0 cycle in deionized water) is 31 12 m. ( ) Number of specimens in each group

PAGE 53

43 By applying equation 4.9 to equation 4.3 4.7, and 3.2, we obtain corrected values for Y, Y'/ Z, and C 0 1/2 In Table 4.5, Y increases from 0.98 0.1 to 1.1 0.1 and C 0 1/2 increases from 0.98 0.1 to 1.1 0.2 for specimens loaded for 10 5 cycles when cal is substituted into equation 4.7 and 3.2. Although the constant values increase after applying the corrected cal into equation 4.3-4.8, the increase is still not significant. The possible reason for the decrease of the geometric constants in this group of specimens that contain the largest and deepest critical flaw size may be the transition from a surface flaw to an internal flaw condition. For example, the Y value decreases from 1.26 for surface flaws to 1.12 for internal flaws. The decrease in the Y value in this study suggests the dominance of internal flaws. However, there were only two specimens in this group that failed in this manner. Therefore, no conclusive explanation can be made to accept or reject the proposed hypothesis that the depth of the crack is responsible for reduction in the constant values. Table 4.5. Geometric factors of glass specimens tested in deionized water with adjusted stress. No. of cycles (Eq.4.3) Y'/Z (Eq.4.5) C 0 1/2 (Eq.4.7) Y (Eq.3.2) corr (Eq.4.3) Y'/Z corr (Eq.4.5) C 0 1/2 (Eq.4.7) Y corr (Eq.3.2) 10 3 (9) 0.24 0.09 1.12 0.17 1.27 0.21 1.27 0.21 0.26 0.09 1.17 0.18 1.32 0.22 1.33 0.22 10 4 (4) 0.19 0.12 1.14 0.25 1.21 0.31 1.21 0.31 0.26 0.15 1.23 0.33 1.29 0.34 1.30 0.32 10 5 (2) 0.15 0.04 0.98 0.18 0.98 0.11 0.98 0.11 0.18 0.04 1.07 0.19 1.08 0.17 1.07 0.12

PAGE 54

44 Based on data in Tables 4.3 and 4.4, the constant values (0.19 0.24) and Z (1.51 1.66) appear to be in the range reported in previous studies (53, 54) with higher SD values. However, the previous investigators did not clearly identify the failure origins of the specimens and the strength degradation analyses were based only on the cone crack geometry that can be determined from fracture mechanics equations. The shape of critical cracks and failure origin sites were not identified. It is important to identify the failure origins; otherwise, one cannot determine the correction factors for crack geometry and free surfaces. Therefore, we can obtain only minimal information from the constant values and Z. In contrast, the constant values C 0 1/2 and Y were determined from the surface area of critical cracks. An assumption was made for the shape of the critical flaws to simplify the calculation methods. For a constant value of C 0 1/2 a triangular crack shape was assumed to calculate the real surface area of the critical crack that propagated from point A-A to A' on a plane of a cone crack. The projected area in the direction of the maximum tensile stress was used to determine the geometric constant value for Mode I crack propagation. For the Y value, we measured the size of a triangular crack and calculated the crack area directly from the fracture surface. The constant values C 0 1/2 and Y are in good agreement for glass specimens. The surface area approach is more rational because the actual size and shape of critical flaws are included in the calculation procedures. These constant values also represent the flaw characteristics, compared with the regular semi-elliptical or semi-circular flaws. However, as discussed earlier regarding the depth of critical cracks, we should consider the surface-to-internal flaw characteristics as the cone crack propagates to a greater depth beneath the contact surface either because of stress corrosion or high load conditions.

PAGE 55

45 Baria-Silicate Glass-Ceramics For baria-silicate glass-ceramics with an aspect ratio of 3/1, we observed the same four patterns of crack configurations (Fig. 4.11 and 4.13) as those observed in glass specimens, i.e., cone-cracks and surface damage cracks. Summarized in Table 4.6 are the crack sizes and geometric factors calculated from equations 4.3-4.8 for group AR3 specimens tested in air. Crack sizes as a function of loading cycles for group AR 3 specimens tested in air and in deionized water are shown in Fig. 4.12. In this group, cone cracks were observed in specimens subjected to cyclic loading in air for 10 4 and 10 5 cycles. No cone crack or contact-induced surface damage was observed in specimens subjected to loading for 10 3 cycles. For specimens subjected to loading at 10 3 cycles, the mean crack size was 57 19 m, which is not statistically significantly different ( p > 0.05 ) from that of the control group. For specimens subjected to cyclic loading at 10 4 and 10 5 cycles in air that failed from a cone crack, Y, Y'/Z, and C 0 1/2 values are greater than those calculated for the glass specimens tested in air and the crack sizes are smaller. One possible reason for the greater constant values is that the cone cracks, which developed in the AR 3 group, were not as deep as those found in the glass group and act as surface cracks. If we consider the shape of the fracture-initiating flaw as a surface crack, then recalculation of a new crack size results in a larger crack dimension. As a result, the Y factor calculated from this new crack size using equation 4.8 is 1.26, and represents the fracture mechanics correction factor for free surfaces (57). Another possible reason is the influence of the local residual stress that develops during cyclic fatigue testing, which one must take into account to calculate the Y value.

PAGE 56

46 For group AR 3 specimens, the fracture toughness calculated from the control group using fractographic analysis is 1.26 0.05 MPam 1/2 The increase in fracture toughness from 0.7 MPam 1/2 in the base glass to 1.26 MPam 1/2 in the AR 3 group is responsible for the increase in the numbers of cycles needed to initiate the cone-crack (from 10 3 cycles to 10 4 cycles). The fracture surfaces of group AR 3 specimens subjected to cyclic loading between 10 3 to 10 5 cycles are shown in Fig. 4.11. For group AR 3 specimens tested in deionized water, cone cracks first appeared in specimens after 10 3 loading cycles, i.e., fewer cycles than required for specimens tested in air. The geometric constant values are similar to those of specimens tested in air. However, the mean critical crack sizes of specimens loaded for 10 4 and 10 5 cycles in deionized water were significantly larger than those in specimens tested in air for the same number of loading cycles. The geometric constant values decrease as a result of increasing crack sizes at a constant stress level. Representative SEM images of fracture surfaces of group AR3 specimens loaded in deionized water are shown in Fig. 4.13. The contact-induced surface damage can be seen on the loaded surfaces of specimens subjected to 10 4 cycles. The contact damage was more severe in specimens subjected to 10 5 cycles. As shown in Fig. 4.13b, a surface impression is noticeable, and the zone of contact damage beneath the loading site is confined within the cone crack. The failure of this specimen did not start from the cone crack, but instead initiated from within the surface contact damage area. More details about the failure initiation process are discussed in the section on fatigue failure mechanisms.

PAGE 57

47 Table 4.6. Crack sizes and geometric factors of group AR 3 specimens tested in air. No. of cycles R (m) (Eq. 4.3) L (m) (Eq.4.5) A p (m) (Eq. 4.7) (ab) 1/2 (m) (Eq. 3.2) Y'/Z (Eq. 4.5) C 0 1/2 (Eq. 4.7) Y (Eq. 3.2) 10 4 (6) 63 23 89 29 96 25 81 21 1.50 0.14 1.34 0.06 1.54 0.11 10 5 (8) 90 53 112 20 123 17 96 10 1.53 0.31 1.28 0.11 1.53 0.11 The mean critical crack size for the control group (0 cycles) is 53 11 m. ( ) Number of specimens in each group Table 4.7. Crack sizes and geometric factors of group AR 3 specimens tested in deionized water. No. of cycles R (m) (Eq. 4.3) L (m) (Eq. 4.5) A p (m) (Eq. 4.7) (ab) 1/2 (m) (Eq. 3.2) Y'/Z (Eq. 4.5) C 0 1/2 (Eq. 4.7) Y (Eq. 3.2) 10 3 (4) 53 14 77 19 80 17 63 17 1.48 0.20 1.44 0.04 1.63 0.05 10 4 (6) 62 12 109 22 118 6 96 5 1.34 0.07 1.23 0.07 1.36 0.07 10 5 (3) 69 9 139 24 140 11 113 6 1.25 0.11 1.22 0.07 1.36 0.09 The mean critical crack size for the control group (0 cycles) is 53 11 m. ( ) Number of specimens in each group

PAGE 58

48 (a) 0 cycles (150X) (b) 10 3 cycles (150X) (c) 10 4 cycles (230X) (d) 10 5 cycles (200X) Figure 4.11. SEM images of fracture surfaces of group AR 3 specimens subjected to cyclic loading from 10 3 to 10 5 cycles: (a) and (b) reveal semi-elliptical cracks and (c) and (d) show triangular cracks.

PAGE 59

49 (a) Air 020406080100120140160010000100000Loading cyclesCrack size (m) R L Ap (ab)1/2 (b) Deionized water 0204060801001201401601800100010000100000Loading cyclesCrack size (m) R L Ap (ab)1/2 Figure 4.12. Crack size as a function of loading cycles for AR 3 specimens. The legends in the graphs refer to the parameters used in Eqs. 3.2, 4.3, 4.5, 4.7 and in Table 4.6 and 4.7.

PAGE 60

50 (a) 10 4 cycles (100X) (b) 10 5 cycles (75X) Figure 4.13. SEM images of fracture surfaces of group AR 3 specimens subjected to cyclic loading for 10 4 and 10 5 cycles in deionized water.

PAGE 61

51 For the AR 8 group, the critical crack sizes are summarized in Table 4.8. Crack sizes as a function of loading cycles for group AR 8 specimens both tested in air and in deionized water are shown in Fig. 4.14. Failure of all specimens initiated from surface damage caused by either polishing or cyclic loading. The mean crack sizes of specimens subjected to cyclic loading in ambient room temperature show no increase after 10 4 to 10 5 cycles, and no contact surface damage, i.e., ring or cone crack was observed in most of the specimens. On the other hand, the strength of these bars tested in deionized water decreased rapidly after 10 3 cycles, corresponding to a considerable increase in critical crack size. In specimens subjected to loading for 10 4 to10 5 cycles in deionized water, a surface impression and subsurface damage zone can be clearly seen. The fracture surfaces of group AR 8 specimens subjected to cyclic loading for 10 3 to 10 5 cycles in air and in deionized water are shown in Figs. 4.15 and 4.16. Table 4.8. Critical crack sizes of fractured group AR 8 specimens. AR 8 0 cycles 10 3 cycles 10 4 cycles 10 5 cycles Air (m) 56 12 56 7 69 13 71 8 Deionized water (m) 92 7 76 16 108 26 265 81

PAGE 62

52 0501001502002503003504000100010000100000Loading cyclesCrack size (m) Air Deionized water Figure 4.14. Crack size as a function of loading cycles for group AR 8 specimens tested in air and in deionized water.

PAGE 63

53 (a) 10 4 cycles (65X) (b) 10 5 cycles (100X) Figure 4.15. SEM images of fracture surfaces of group AR8 specimens subjected to loading for 10 4 and 10 5 cycles.

PAGE 64

54 (a) 10 4 cycles (200X) (b) 10 5 cycles 120X Figure 4.16. SEM images of fracture surfaces of group AR 8 specimens subjected to loading for 10 4 and 10 5 cycles in deionized water.

PAGE 65

55 Fatigue Failure Mechanisms Representative SEM images of fracture surfaces for baria-silicate glass-ceramics subjected to loading for 10 5 cycles in deionized water for each aspect ratio group are shown in Figs. 4.17 (AR 3) and 4.21 (AR 8). Figure 4.17 shows the fracture surfaces and critical crack size for an AR 3 specimen (aspect ratio 3/1). The fracture surface appears to be tortuous and boundaries of crystallites can be clearly seen. Contact damage can be seen at the top surface. A crushing zone occurred near the surface, including microcracks propagating downward from the contact surface. A ring crack developed outside the contact circle and propagated downward to form a cone crack. However, the failure of this specimen did not start from the tip of a cone crack, as occurred in the AR 3 specimens tested in air, where the failure initiation sites were usually associated with the cone crack without evidence of severe contact damage (Fig. 4.19a). For AR 3 specimens tested in air, slight roughening of the contact surface of specimens subjected to 10 5 loading cycles was the only sign indicating damage from cyclic loading in addition to a ring and cone crack formation. For AR 3 specimens tested in deionized water, an area of microcracks propagating downward from the contact surface can be clearly seen in Fig. 4.18b. This contact damage zone was a failure initiation site for catastrophic fracture as shown in Fig. 4.13b. The aqueous environment caused a change in the fracture initiation site from the tip of a cone crack to surface contact damage by accelerating the crack initiation and propagation rate at the contact surface. The wear surface was first observed in group AR 3 specimens subjected to 10 4 cycles in deionized water. The damage was more severe for specimens loaded for 10 5 cycles as shown in Fig. 4.19. Two principal factors are associated with the formation of microcracks within the surface and subsurface damage zone, stress corrosion fatigue and residual stress. Stress corrosion

PAGE 66

56 fatigue can easily occur at the surface damage area during cyclic loading because there is much material exposed to the environment and this process enhances the propagation of the microcracks. The formation of these microcracks also increases the compliance of the material in the damage zone. As a result, the residual stress develops from a mismatch between the elastic modulus of the damage zone and the surrounding area. This residual stress also enhances the formation of microcracks within the contact damage area. The strain mismatch between the glass and crystal phases associated with cyclic loading can also cause residual tensile stress during the unloading half-cycle (18). Hill (58) observed the fracture path of group AR 3 specimens propagating through the glass matrix. He suggested that crack deflection is a toughening mechanism for this material. Failure of specimens used in a previous study (58) occurred during a quasi-static four-point loading test. Therefore, failure origins of those specimens were caused by flaws produced during grinding and polishing. In the present study the critical crack propagated from a cone crack that developed during blunt indentation loading, or that originated from the damage zone beneath the contact surface. The propagation path for the cone crack appeared to progress straight downward and likely through the crystal phase (Fig.4.18a), following the path perpendicular to the direction of the maximum tensile stress. Therefore, a transgranular fracture mode was observed after cyclic loading in group AR 3 specimens. Analytical results from previous studies suggest that failure from a cone crack should initiate from one of the two opposite points on the base rim of the cone (53-54). However, we found that only a few specimens failed from the tip of the cone crack. Instead, most of the AR 3 specimens failed from a triangular crack developed during fast fracture.

PAGE 67

57 Critical crack Figure 4.17. Montage of SEM images (1000X) of fracture surfaces of group AR 3 specimen subjected to cyclic loading for 105 cycles in deionized water: Upper left 100X; Upper right and Bottom 1000X.

PAGE 68

58 Figure 4.21 illustrates the fracture surfaces of the AR 8 group (aspect ratio 8/1). The surfaces appear to be more tortuous, but with sharp deflection angles instead of the smoother curves seen in the fracture path for the AR 3 group (Fig. 4.22a). Hill (58) suggested that the crack path of this group was transgranular through the baria-silicate crystals indicating that the fracture propagated along a new path of least resistance rather than around the crystals as seen in the AR 3 group. He also suggested crack bridging as an additional toughening mechanism in addition to crack deflection for this group. Cone crack Microcracks (a) (b) Figure 4.18. SEM image of fracture surfaces of group AR3 specimens subjected to loading for 10 5 cycles in deionized water (2000X): (a) cone crack-related failure (higher magnification of Fig. 4.13a); (b) surface damage-related failure (higher magnification of Fig. 4.17).

PAGE 69

59 Ring crack (a) Contact dam a ge (b) Figure 4.19. SEM images of surface contact damage in AR 3 specimens: (a) after 10 5 cycles in air; (b) after 10 5 cycles in deionized water.

PAGE 70

60 In the present study, transgranular fracture occurred through the crystal phase in the crushing zone beneath the contact surface and other areas (Fig.4.21), suggesting the same fracture path and the toughening mechanism as that which occurred in the quasi-static loading failure. We did not observe failure originating from a cone crack in AR 8 specimens. Based on the fracture surface analysis of group AR 8 specimens tested in air, the mean crack size increased slightly in proportion to the increase in numbers of loading cycles. However, no subsurface damage was detected in those specimens except for the slight rough surface observed after 10 5 cycles (Fig.4.20a). In contrast, group AR 8 specimens tested in deionized water showed a significant effect of stress corrosion fatigue (Fig. 4.20b). The mean strength values of these bars tested in deionized water decreased rapidly after 10 3 cycles, corresponding to a significant increase in critical crack sizes. In specimens subjected to loading for 10 4 to10 5 cycles (Fig. 4.21) in deionized water, a surface impression and subsurface damage zone can be clearly seen. Areas of microcracks propagating downward from the contact surface were observed and are shown in Fig. 4.22. The dominant fracture path is transgranular (Fig. 4.22a), which is similar to that observed after quasi-static loading failure (58). The mechanisms of microcrack formation in group AR 8 specimens were similar to those described earlier for group AR 3 specimens. However, the effect of an aqueous environment was significant for group AR 8 specimens. Because of its high fracture toughness (2.11 0.16 MPam 1/2 ), group AR 8 specimens are more resistant to crack propagation compared with the glass and AR 3 groups, which we can verify for the specimens tested in air. However, the aqueous environment accelerated the crack propagation rate at the surface contact area in both AR 3 and AR 8 groups. Thus, we

PAGE 71

61 observed severe surface damage in specimens tested in deionized water. This resulted in a significant decrease in the mean strength for group AR 8 specimens because the failure initiation site for group AR 8 specimens was at the contact surface. For group AR 3 specimens, even though we observed severe surface damage, the failure initiation sites of specimens tested in air were different from those observed in specimens tested in deionized water. Because of the change of failure initiation sites from a cone-crack related flaw to contact surface damage, we did not observe a significant decrease in strength for group AR 3 specimens tested in air and in deionized water after 10 5 cycles. In this study, the contact fatigue damage patterns of glass-ceramic specimens are similar to the wear process of silicon nitride observed in a previous study (43). The wear mechanisms of silicon nitride with a fine surface finish can be divided into three stages. In stage I, roughening of the contact surface occurs because of the damage to the grain boundaries. In stage II, ring and cone cracks form and propagate. In stage III, a significant amount of material is lost from the region outside the contact area. If the surface finish is poor, stage II can occur before stage I (43). Stage I occurs because of the viscoelastic behavior of the glass phase that is distributed along the grain boundaries of silicon nitride during the sintering process. The delayed response of the glass phase allows tensile stress to develop during unloading because of the strain mismatch between the two phases. Eventually, the cyclic tensile stress in the glass phase causes grain boundary fracture. The delayed response of the glass phase as well as the large loads also causes shear stress to develop just outside the contact region. This distribution of shear stress can enhance crack initiation near the contact circle. In this study, roughening of the contact surface was observed on

PAGE 72

62 specimens subjected to cyclic loading in group AR 3 and group AR 8 specimens tested in deionized water (Fig.4.23). The glass matrix in baria-silicate glass-ceramics behaves as a site for the residual tensile stress to occur. The protruding crystals and sites associated with lost crystals are clearly seen in Fig. 4.23. Stage II occurs because of the high tensile stress at the edge of the contact region in the Hertzian stress field and the existence of the surface cracks. In this study, a ring and cone crack were observed in most of the AR 3 specimens subjected to a sufficient number of loading cycles (Fig. 4.19). For group AR 8 specimens, no ring or cone crack was observed because of the dissipation of the contact stress that can decrease the intensity of the Hertzian stress fields (Fig. 4.20). Stage III involves the removal of material from the surface and may resemble the spallation caused by subsurface cracks that propagate to the surface (59). For stage III wear, cracks start at the contact surface and propagate downward a certain distance and then progress upwards toward the surface (43, 59). The reason for the change in the direction of the propagating cracks may be the change in stress fields when there is the presence of cracks in the material structure. High tensile stress fields are observed close to the crack tip (59). A high tensile stress field at the surface and at the crack tip can cause the cracks to propagate toward the surface. The upward-curving cracks eventually result in a loss of material from the region outside the contact region. In the present study, no stage III wear was observed in the groups AR 3 and AR 8 tested in air. In contrast, severe surface damage caused from material loss was observed in group AR 3 and AR 8 specimens tested in deionized water (Fig.4.24). Thus, stress corrosion has a significant effect at this stage, by increasing the rate of the crack initiation and

PAGE 73

63 propagation in baria-silicate glass-ceramics. This process results in the creation of strength-limiting flaws that cause significant decreases in the strength of glass-ceramic materials. In summary, cyclic loading creates strength-limiting flaws that can affect the fracture susceptibility of ceramic materials. For materials with low toughness (glass and AR 3), a ring and cone crack were the dominant flaws that caused the decrease in the mean strength of these materials. For the material with a higher toughness (AR 8), surface flaws created during contact fatigue were strength limiting. Stress corrosion enhanced the crack propagation rate and caused a significant decrease in strength of the AR 8 group. The contact fatigue damage patterns observed in this study were similar to the wear process reported in a previous study (43), including three stages of progressive surface and subsurface damage. In this study, the glass-ceramics were produced with two aspect ratios using controlled heat treatment regimens to study the effect of microstructure on fatigue failure mechanisms. We selected a baria-silicate composition because there should be little residual stress caused by elastic mismatch or a thermal expansion mismatch during the cooling process. Therefore, we minimized the effect of the residual stress caused from these mismatches and we observed the actual cyclic behavior of baria-silicate glass-ceramics. Previous studies showed that the microstructure of a micaceous glass-ceramic system was a major factor in controlling the damage modes under Hertzian contact loading (9, 10). By heat-treating glass specimens, the investigations yielded a series of glass-ceramics with microstructures that contained fine (0.3 m x 1 m) and coarse

PAGE 74

64 platelets (1.2 m x 8 m). Contact damage patterns from blunt indentation of bonded-interface specimens can be classified into two damage modes: (1) ring and cone cracks that are dominant in fine microstructures, and (2) subsurface deformation or quasi-plasticity that occurred in coarse microstructures. The authors concluded that microstructural coarsening of micaceous glass-ceramics produced a brittle-plastic transition. (a) Contact damage (b) Contact damage Figure 4.20. Images of surface contact damage in group AR 8 specimens (a) after 10 5 cycles in air; (b) after 10 5 cycles in deionized water.

PAGE 75

65 Critical crack Figure 4.21. Montage of SEM images of fracture surfaces of group AR 8 specimens subjected to 105 cycles (1000X) in deionized water: Upper left 100X; Upper right and Bottom 1000X.

PAGE 76

66 Transgranular fracture (a) 10 4 cycles Transgranular fracture C C r r u u s s h h i i n n g g z z o o n n e e (b) 10 5 cycles Figure 4.22. SEM images of fracture surfaces of group AR8 specimens subjected to loading for 10 4 and 10 5 cycles in deionized water (3000X).

PAGE 77

67 Protruding crystals (a) Protruding crystals (b) Figure 4.23. SEM images of the cyclically loaded surfaces of baria-silicate glass-ceramics (a) AR 3 and; (b) AR 8 specimens tested in deionized water.

PAGE 78

68 (a) (b) Figure 4.24. SEM images of the cyclically loaded surfaces of baria-silicate glass-ceramics (a) AR 3 and; (b) AR 8 specimens tested in deionized water.

PAGE 79

69 In the present study, we observed the same fracture patterns as reported earlier, but some additional explanation is necessary. Relative to the critical crack configurations, we observed that the fracture paths progressed through the contact damage in specimens with a high aspect ratio (AR 8), or near the ring crack in glass specimens in a similar manner as was reported in other studies (9, 10). Previous investigators (9, 10) did not identify whether the failure origins occurred from the cone cracks or from contact damage. Based on fractographic analysis of our specimens, we found three different crack configurations that cannot be clearly described by a simple cone-crack model (53). The second point is that the authors (12, 13, 60) also suggested that the Hertzian test simulates oral loading conditions by using contact loads that are similar to intra-oral occlusal loads, and an indenter radius that is comparable to the cuspal radii of teeth. Indentation loads of up to 3000 N were used, which were very high compared to normal occlusal loads (89-890 N) (61). Furthermore, for indented specimens, the tungsten carbide indenters that they used have a much higher elastic modulus than that of tooth enamel. As a result, the authors suggested that a quasi-plastic zone beneath the contact area was a failure initiation site for alumina and silicon carbide materials (27). This quasi-plastic zone developed in a compressive stress field caused from Hertzian contact loading. In the present study, the cyclic contact load of 200 N was low compared with that used in the previous studies. The damage zone was confined near the surface. It is evident that microcracks formed in that damage zone originated and propagated downward from the contact surface, not from the compressive zone beneath the contact area.

PAGE 80

70 Clinical failure mechanisms are currently a topic of a major concern. Although some investigators prefer a blunt indentation test since it can characterize materials properties under clinically relevant conditions, others argue that indentation damage is typically not associated with clinical failures of dental prostheses (9-12, 62, 63). From a fractographic perspective, failure initiation sites of failed ceramic prostheses appear to occur at the cementation surface or the interface between core and veneer (14, 15), not within occlusal surface. In spite of this difference, there is little doubt that contact damage can occur during mastication, because wear of the occlusal surfaces can be generally observed in ceramic prostheses. Even though this damage may not lead to immediate failure, a decrease in the critical stress for failure can be expected. A partial loss of ceramic prosthesis structure (chipping) caused by contact damage is also possible as a treatment failure because of compromised esthetics and function. To obtain useful in vitro information that is relevant to clinical failure conditions, well-designed experiments under variable loading parameters and simulated clinical conditions are recommended for future studies of the failure behavior of dental ceramics.

PAGE 81

CHAPTER 5 CONCLUSIONS The conclusions of this study are as follows: (1) Cyclic loading decreases the strength of baria-silicate glass and glass-ceramics. (2) Four different crack geometries were identified and associated with failure mechanisms. For glass specimens, the decrease in fracture stress after 10 3 cycles was associated with cone crack formation. No significant additional reduction was found after loading for 10 3 cycles and 10 5 cycles of specimens tested both in air and in deionized water. (3) For the glass-ceramic groups with a crystal aspect ratio of 3/1 tested in air, the decrease in fracture stress occurred initially after 10 4 cycles and was associated with the formation of cone cracks. The decrease in fracture stress observed after 10 3 cycles for specimens tested in deionized water, suggests an acceleration of crack propagation because of stress corrosion. (4) For the glass-ceramic groups with a crystal aspect ratio of 8/1, we did not observe failure originating from a cone crack in any specimens. We observed a gradual decrease in fracture stress in specimens subjected to loading in air for 10 3 to 10 5 cycles. A statistically significant decrease was found in specimens cyclically tested in deionized water as compared to those tested in air. A reduction of approximately 50% in fracture stress was found for specimens subjected to loading for 10 5 cycles in deionized water. (5) Comparisons between methods used to determine the possible crack sizes were made. The geometric constants obtained from the surface area analyses (Equations 3.2 71

PAGE 82

72 and 4.7) are more rational than those obtained from the cone-crack analysis models (Equations 4.3 and 4.5). However, an error is associated with the methods that are used to determine the surface area of critical cracks in this study because the curvature of a cone crack was disregarded to simplify the calculation. (6) Control of microstructure by heat treatment can greatly affect the strength and failure mechanisms of glass-ceramics during cyclic fatigue testing in air. The failure initiation mechanisms change from one occurring at the base of the cone-crack in baria-silicate glass and glass-ceramics with a crystal aspect ratio of 3/1 (AP 3) to that resulting from surface contact damage in glass-ceramics with an crystal aspect ratio of 8/1 (AR 8). (7) An intergranular fracture path was observed in glass-ceramics with an aspect ratio of 3/1, suggesting that the crack deflection is a major toughening mechanism. For an aspect ratio of 8/1, a transgranular fracture mode was dominant and crack bridging and crack deflection contributed to toughening. (8) Stress corrosion fatigue accelerated the crack propagation rate in baria-silicate glass-ceramic specimens, especially in specimens with an aspect ratio of 8/1. The mean crack sizes were significantly larger in AR 8 specimens subjected to cyclic loading in deionized water, compared with those subjected to the same numbers of cycles in air.

PAGE 83

REFERENCES 1. Walton JN, Gardner FM, Agar JR (1986). A survey of crown and fixed partial denture failures: Length of service and reasons for replacement. J Prosthet Dent 56:416-21. 2. Foster LV (1990). Failed conventional bridge work from general dental practice: Clinical aspects and treatment needs 142 cases. Br Dent J 168:199-201. 3. Moffa JP, Lugassy AA, Ellison JA (1988). Clinical evaluation of a castable ceramic material (abstract). J Dent Res 67:118. 4. White SN (1993). Mechanical fatigue of a feldspathic dental porcelain. Dent Mater 9:260-4. 5. White SN, Zhao XU, Yu Z, Li ZC (1995). Cyclic mechanical fatigue of a feldspathic dental porcelain. Int J Prosthodent 8:1-8. 6. White SN, Li ZC, Yu Z, Kipnis V (1997). Relationship between static chemical and cyclic mechanical fatigue in a feldspathic porcelain. Dent Mater 13:103-10. 7. Sobrinho LC, Cattell MJ, Glover RH, Knowles JC (1998). Investigation of the dry and wet fatigue properties of three all-ceramic crown systems. Int J Prosthodont 11:255-62. 8. Morena R, Beaudreau GM, Lockwood PE, Evans AL, Fairhurst CW (1986). Fatigue of dental ceramics in a simulated oral environment. J Dent Res 65:993-7. 9. Peterson IM, Wuttiphan S, Lawn BR, Chyung K (1998). Role of microstructure on contact damage and strength degradation of micaceous glass-ceramics. Dent Mater 14:80-9. 10. Peterson IM, Parjares A, Lawn BR, Thompson VP, Rekow ED (1998). Mechanical characterization of dental ceramics by Hertzian contacts. J Dent Res 77:589-602. 11. Cai H, Kalceff MS, Lawn BR (1994). Deformation and fracture of mica-containing glass-ceramics in Hertzian contacts. J Mater Res 9:762-70. 12. Cai H, Kalceff MS, Hooks BM, Lawn BR (1994). Cyclic fatigue of a mica-containing glass-ceramics at Hertzian contacts. J Mater Res 9:2654-61. 73

PAGE 84

74 13. Lawn BR, Deng Y, Thompson VP (2001). Use of contact testing in the characterization and design of all-ceramic crownlike layer structures: a review. J Prosthet Dent 86:495-510. 14. Kelly JR, Giordano R, Pober R, Cima MJ (1990). Fracture surface analysis of dental ceramics: clinically failed restorations. Int J Prosthodont 3:430-40. 15. Thompson JY, Anusavice KJ, Naman A, Morris HF (1994). Fracture surface characterization of clinically failed all-ceramic crowns. J Dent Res 73:1824-32. 16. MacDowell JF (1965). Nucleation and crystallization of barium silicate glasses. Proc Brit Ceram Soc 3:229-40. 17. Freiman SW, Onoda GY, Pincus AG (1974). Mechanical properties of 3BaO-5SiO 2 glass. J Am Ceram Soc 57:8-12. 18. Suresh S (1998). Fatigue of materials. Cambridge University Press, Cambridge, U.K. 19. Ritchie RO (1988). Mechanisms of fatigue crack propagation in metals, ceramics and composites: role of crack tip shielding. Mater Sci Eng A103:15-28 20. Ritchie RO, Dauskardt RH, Yu W (1990). Cyclic fatigue-crack propagation, stress-corrosion, and fracture-toughness behavior in pyrolytic carbon-coated graphite for prosthetic heart valve applications. J Biomed Mater Res 24:189-206. 21. Ritchie RO, Gilbert CJ, McNaney JM (2000). Mechanics and mechanisms of fatigue damage and crack growth in advanced materials. Int J Solids Struc 37:311-29. 22. Lathabai S, Rodel J, Lawn BR (1991). Cyclic fatigue from frictional degradation at bridging grains in alumina. J Am Ceram Soc 74:1340-8. 23. Dauskardt RH (1993). A frictional wear mechanism for fatigue-crack growth in grain bridging ceramics. Acta Metal Mater 41:2765-81. 24. Lawn BR, Padture NP, Guiberteau F, Cai H (1994). A model for microcrack initiation and propagation beneath Hertzian contacts in polycrystalline ceramics. Acta Metal Mater 42:1683-93. 25. Argon AS (1982). Fracture in compression of brittle solids. National Materials Advisory Board Report NMAB-404, National Academy Press, Washington, DC. 26. Ewart R, Suresh S (1987). Crack propagation in ceramics under cyclic loads. J Mater Sci 22:1173-92. 27. Lawn BR (1998). Indentation of ceramics with spheres: a century after Hertz. J Am Ceram Soc 81:1977-94.

PAGE 85

75 28. Roesler FC (1956). Brittle fractures near equilibrium. Proc Phys Soc London B69:981-92. 29. Lawn B (1993). Fracture of brittle solids. Cambridge University Press, Cambridge, U.K. 30. Lee SK, Lawn BR (1998). Role of microstructure in Hertzian contact damage in silicon nitride: strength degradation. J Am Ceram Soc 81:997-1003. 31. Wiederhorn SM, Bolz LH (1970). Stress corrosion and static fatigue of glass. J Am Ceram Soc 53:543-8. 32. Michalske TA, Freiman SW (1982). A molecular interpretation of stress corrosion in silica. Nature 295:511-2. 33. Ebrahimi ME, Chevalier J, Fantozzi G (1999). Slow crack-growth behavior of alumina ceramics. J Mater Res 15:142-7. 34. Chevalier J, Olagnon C, Fantozzi G (1999). Subcritical crack propagation in 3Y-TZP ceramics: static and cyclic fatigue. J Am Ceram Soc 82:3129-38. 35. Ebrahimi ME, Chevalier J, Fantozzi G (2000). Environmental effects on crack propagation of alumina ceramics. Ceram Eng Sci Proc 21:385-96. 36. Asoo B, McNaney JM, Mitamura Y, Ritchie RO (2000). Cyclic fatigue-crack propagation in sapphire in air and simulated physiological environments. J Biomed Mater Res 52:488-91. 37. Tanaka H, Bando Y (1990). Atomic crack tips in silicon carbide and silicon crystals. J Am Ceram Soc 73:761-3. 38. Cook RH (1986). Crack propagation thresholds: a measure of surface energy. J Mater Res 1:852-60. 39. Thomson R (1990). The molecular wedge in a brittle crack: a simulation of mica/water. J Mater Res 5:524-34. 40. Hertzberg RW (1995). Deformation and fracture mechanics of engineering materials. John Wiley & Sons, Inc., New York, U.S.A. 41. Choi G, Horibe S (1995). The environmental effect on cyclic fatigue behavior in ceramic materials. J Mater Sci 30:1565-9. 42. Jacobs DS, Chen IW (1995). Cyclic fatigue in ceramics: a balance between crack shielding accumulation and degradation. J Am Ceram Soc 78:513-20. 43. Chen Z, Cuneo JC, Mecholsky JJ, Hu S (1996). Damage processes in Si 3 N 4 bearing material under contact loading. Wear 198:197-207.

PAGE 86

76 44. Paris PC, Gomez MP, Anderson WP (1961). A rational analytic trend of fatigue. Trend in Engineering 13:9-14. 45. Sylva LA, Suresh S (1989). Crack growth in transforming ceramics under cyclic tensile loads. J Mater Sci 24:1729-38. 46. Dauskardt RH, Marshall DB, Ritchie RO (1990). Cyclic fatigue-crack propagation in magnesia-partially-stabilized zirconia ceramics. J Am Ceram Soc 73:893-903. 47. Jones DW, Sutow EJ (1987). Stress corrosion failure of dental porcelain. Br Ceram Trans J 86:40-3. 48. Adam R, McMillan PW (1977). Static fatigue in glasses. J Mat Sci 12:643-57. 49. Sherrill CA, OBrien WJ (1974). Transverse strength of aluminous and feldspathic porcelain. J Dent Res 53:683-90. 50. Southan DE, Jorgensen KD (1974). The endurance limit of dental porcelain. Aust Dent J 19:7-11. 51. Irwin GR (1962). Crack-extension force for a part-through crack in a plate. J Appli Mech 29:651. 52. Mecholsky JJ, Passoja DE, Feinberg-Ringel KS (1989). Quantitative analysis of brittle fracture surfaces using fractal geometry. J Am Ceram Soc 72:460-5. 53. Lawn BR, Wiederhorn SM, Johnson HH (1975). Strength degradation of brittle surfaces: blunt indenters. J Am Ceram Soc 58:428-32. 54. Evans AG (1973). Strength degradation by projectile impacts. J Am Ceram Soc 56:405-9. 55. Murakami Y (1985). Analysis of stress intensity factors of mode I, II and III for inclined surface cracks of arbitrary shape. Eng Frac Mech 22:101-14. 56. Widjaja S, Ritter JE, Jakus K (1996). Influence of R-curve behavior on strength degradation due to Hertzian indentation. J Mater Sci 31: 2379-84. 57. Brown WF, Strawley JE (1967). Plane strain crack toughness testing of high strength metallic materials. ASTM Special technical publication No. 410: 89-129. 58. Hill TJ, Mecholsky JJ, Anusavice KJ (1995). Fracture toughness versus crystallization in baria-silicate glass-ceramics. Am Assoc Dent Res Abstract 1657. 59. Chao L, Shetty DK, Adair JH, Mecholsky JJ (1996). Development of silicon nitride for rolling-contact bearing applications a review. J Mater Edu 17:245. 60. Jung YG, Peterson IM, Kim DK, Lawn BR (2000). Lifetime-limiting strength degradation from contact fatigue in dental ceramics. J Dent Res 79:722-31.

PAGE 87

77 61. Anusavice KJ (1996). Philllips science of dental materials, 10 th edition. W.B. Saunders Company, Philadelphia, USA. 62. Kelly JR, Tesk JA, Sorensen JA (1995). Failure of all-ceramic fixed partial dentures in vitro and in vivo: analysis and modeling. J Dent Res 74:1253-8. 63. Kelly JR (1999). Clinically relevant approach to failure testing of all-ceramic restorations. J Prosthet Dent 81:652-61.

PAGE 88

BIOGRAPHICAL SKETCH Kallaya Suputtamongkol was born in Bangkok, Thailand, on December 15, 1968. After graduation from a local high school in 1984, she entered Mahidol University and earned a Doctor of Dental Surgery degree in 1990. She continued her graduate study at Mahidol University and obtained a Master of Science degree in prosthodontics in 1997. After graduation, she has started working at the Faculty of Dentistry, Mahidol University, as an instructor in the Prosthodontics Department. She has continued her education by entering the Ph.D program at the University of Florida, College of Engineering, in 1998. 78


Permanent Link: http://ufdc.ufl.edu/UFE0000800/00001

Material Information

Title: Contact Fatigue Mechanisms as a Function of Crystal Aspect Ratio in Baria-Silicate Glass-Ceramics
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000800:00001

Permanent Link: http://ufdc.ufl.edu/UFE0000800/00001

Material Information

Title: Contact Fatigue Mechanisms as a Function of Crystal Aspect Ratio in Baria-Silicate Glass-Ceramics
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000800:00001


This item has the following downloads:


Full Text












CONTACT FATIGUE MECHANISMS
AS A FUNCTION OF CRYSTAL ASPECT RATIO
IN BARIA-SILICATE GLASS-CERAMICS














By

KALLAYA SUPUTTAMONGKOL


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2003







































Copyright 2003

by

Kallaya Suputtamongkol















ACKNOWLEDGMENTS

I would like to thank my advisors, Dr. K.J. Anusavice and Dr. J.J. Mecholsky Jr. I

greatly appreciate the support and guidance they have provided for during my studies

here at UF. They have also taught me valuable lessons of good research. I have had an

excellent learning experience, which I can use to develop my career skills as a teacher.

Special thanks go to Ben Lee who has always been there to help. Without Ben, I could

not have gone home as soon as is now possible. Very special thanks go to my mother

who has always encouraged me. I also would like to thank all my friends for their

support and friendship. They have made my life here a lot easier and more pleasant.
















TABLE OF CONTENTS


A C K N O W L E D G M E N T S ................................................................................................. iii

LIST OF TABLES ..................................................... vi

LIST OF FIGURES ............................. ............ .................................... vii

ABSTRACT .................................................... ................. ix

CHAPTER

1 IN TR O D U C T IO N ................................................................. .... ..... ...............

2 LITER A TU R E R EV IEW .................................................................... ...............4...

Mechanisms of Fatigue Crack Growth in Ceramics...............................................4...
M echanism s of Crack-Tip Shielding ....................................................... ...............6...
C o n ta ct F atig u e ............................... ....................................... ..................... 9
Strength Degradation from Contact Damage ........................................ ................ 12
E nv iron m mental E effect .................................................................................................. 12
Fatigue Behavior of Dental Glass-Ceram ics ......................................... ................ 15
M echanism of Fatigue D am age ..................................... ..................... ............... 18
S p ecific A im s.............................................................................................................. 2 0

3 M A TERIAL S AN D M ETH OD S .......................................................... ................ 21

Preparation of Baria-Silicate Glass-Ceramic Specimens ......................................21
H eat Treatm ent Schedules .................. ............................................................. 21
M icrostructural A analysis ................... .............................................................. 21
Surface Preparation ........................ ........... .......... ...............22
D en sity .................. ............ ...... .............................................................. ........ .. 2 2
Y oung's M odulus and Poisson's R atio.................................................. ................ 22
T e st M eth o d s .............................................................................................................. 2 3
M echanical F atigue T est....................................... ....................... ................ 23
Strength D egradation T est.............................................................. ................ 23
Fractographic A analysis ................ ............. ............................................. 24
Statistical A n aly sis...................................................................................................... 2 5









4 RE SU LTS AN D D ISCU SSION ............................................................ ................ 26

Strength D egradation ... ....................................................................... ................ 29
Fractographic A analysis ................ ............. ............................................. 30
B aria-Silicate G lass ........................................................ ... .. ...... ...... .. ....... ..... 30
Strength Degradation Associated with a Cone-Crack....................................35
B aria-Silicate G lass-Ceram ics.................. .................................................. 45
F atigue F failure M echanism s........................................ ....................... ................ 55

5 C O N C L U SIO N S .................................................. .............................................. 7 1

REFERENCES .................................... ........... .....................................73

BIO GR A PH ICAL SK ETCH .................................................................... ................ 78








































v
















LIST OF TABLES


Table page

2.1. Possible mechanisms of cyclic fatigue crack propagation in ceramics...................7...

3.1. H eat treatm ent schedules ......................................... ........................ ................ 2 1

4.1. Physical properties of baria-silicate glass-ceramics............................ ................ 26

4.2. Mean flexural strength and standard deviation for baria-silicate glass-ceramics. ...28

4.3. Crack sizes and geometric factors for glass calculated from equations 3.2 and
equ atio n s 4 .3 -4 .7 ..................................................................................................... 3 8

4.4. Crack sizes and geometric factors for glass specimens tested in deionized
w after ...................................................................................................... ....... .. 4 2

4.5. Geometric factors of glass specimens tested in deionized water with adjusted
store ss ...................................................................................................... ....... .. 4 3

4.6. Crack sizes and geometric factors of group AR 3 specimens tested in air. ............47

4.7. Crack sizes and geometric factors of group AR 3 specimens tested in deionized
w after ...................................................................................................... ....... .. 4 7

4.8. Critical crack sizes of fractured group AR 8 specimens. ...................................51















LIST OF FIGURES


Figure page

2.1. Schematic illustration of the fatigue crack propagation stages.............................6...

2 .2 H ertzian contact loading ......................................... ......................... .............. 10

2.3. C one crack param eters .................................................................. ............... 13

2.4 Schematic illustration of the reaction of water and a strained Si-O-Si bond at
th e cra ck tip ............................................................................................................. 14

2.5. Schematic illustration of crack propagation in the presence of a chemical
w e d g e ............................................................................................................... . 1 6

3.1. Schem atic illustration of cyclic fatigue fixture. .................................. ................ 24

3.2. Schematic representation of fracture surface features: ai and bi are original
crack depth and width; and acr and bcr are critical crack depth and width. ............25

4.1. SEM images of baria-silicate glass-ceramic microstructures with aspect ratios
of 3/1 (a) and 8/1 (b) after etching with 1% HF for 10 s (4000X)........................27

4.2. Mean strength of baria-silicate glass-ceramics as a function of number of
lo ad in g cy cle s ........................................................................................................... 2 9

4.3. Schematic illustrations of selected failure origins from cone cracks in glass and
glass-ceram ic specim ens (Top view ) .................................................. ................ 31

4.4. SEM images of a cone crack formed in deionized water after 104 loading
cycles. Fracture typically originated from either point B...................................32

4.5. SEM images of a triangular crack in a glass specimen caused by cyclic loading
in air for 105 cycles. Fracture initiated from points A-A ............... ..................... 33

4.6. SEM images of a semi-elliptical surface crack in a glass specimen caused by
cyclic loading in deionized water for 105 cycles................................. ................ 34

4.7. SEM images of a typical semi-elliptical crack in a glass specimen caused by
p o lish in g ............................................................................................................... ... 3 4

4.8. Schematic illustration of a cone crack induced by a Hertzian contact load.............36









4.9. Fracture origin associated with a Hertzian cone crack as suggested by Evans........36

4.10. Crack sizes as a function of cyclic loading for glass specimens: (a) Tested in
air; (b) Tested in deionized w ater ....................................................... ................ 39

4.11. SEM images of fracture surfaces of group AR 3 specimens subjected to cyclic
loading from 103 to 105 cycles: (a) and (b) reveal semi-elliptical cracks and
(c) and (d) show triangular cracks....................................................... ................ 48

4.12. Crack size as a function of loading cycles for AR 3 specimens...............................49

4.13. SEM images of fracture surfaces of group AR 3 specimens subjected to cyclic
loading for 104 and 105 cycles in deionized water. ............................. ................ 50

4.14. Crack size as a function of loading cycles for group AR 8 specimens tested in
air and in deionized w ater ....................................... ........................ ................ 52

4.15. SEM images of fracture surfaces of group AR8 specimens subjected to
loading for 104 and 105 cycles ............................................................. ............... 53

4.16. SEM images of fracture surfaces of group AR 8 specimens subjected to
loading for 104 and 105 cycles in deionized water. ............................. ................ 54

4.17. Montage of SEM images (1000X) of fracture surfaces of group AR 3 specimen
subjected to cyclic loading for 105 cycles in deionized water: ......................... 57

4.18. SEM image of fracture surfaces of group AR3 specimens subjected to loading
for 105 cycles in deionized water (2000X): (a) cone crack-related failure
(higher magnification of Fig. 4.13a); (b) surface damage-related failure (higher
m agnification of F ig 4 .17) ...................................... ....................... ................ 58

4.19. SEM images of surface contact damage in AR 3 specimens: (a) after 105 cycles
in air; (b) after 105 cycles in deionized w ater...................................... ................ 59

4.20. Images of surface contact damage in group AR 8 specimens (a) after 105 cycles
in air; (b) after 105 cycles in deionized w ater...................................... ................ 64

4.21. Montage of SEM images of fracture surfaces of group AR 8 specimens
subjected to 105 cycles (1000X) in deionized water: ........................................65

4.22. SEM images of fracture surfaces of group AR8 specimens subjected to loading
for 104 and 105 cycles in deionized water (3000X)............................................ 66

4.23. SEM images of the cyclically loaded surfaces of baria-silicate glass-ceramics
(a) AR 3 and; (b) AR 8 specimens tested in deionized water...............................67

4.24. SEM images of the cyclically loaded surfaces of baria-silicate glass-ceramics
(a) AR 3 and; (b) AR 8 specimens tested in deionized water...............................68














Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

CONTACT FATIGUE MECHANISMS
AS A FUNCTION OF CRYSTAL ASPECT RATIO
IN BARIA-SILICATE GLASS-CERAMICS

By

Kallaya Suputtamongkol

May 2003

Chair: Kenneth J. Anusavice
Cochair: John J. Mecholsky Jr.
Major Department: Materials Science and Engineering

Ceramic materials are potentially useful for dental applications because of their

esthetic potential and biocompatibility. However, the existence of fatigue damage in

ceramics raises considerable concern regarding its effect on the life prediction of dental

prostheses. During normal mastication, dental restorations are subjected to repeated

loading more than a thousand times per day and relatively high clinical failure rates for

ceramic prostheses have been reported. To simulate the intraoral loads, Hertzian

indentation loading was used in this study to characterize the fatigue failure mechanisms

of ceramic materials using clinically relevant parameters. The baria-silicate system was

chosen because of the nearly identical composition between the crystal and the glass

matrix. Little or no residual stress is expected from the elastic modulus and thermal

expansion mismatches between the two phases. Crystallites with different aspect ratios

can also be produced by controlled heat treatment schedules. The objective of this study









was to characterize the effect of crystal morphology on the fatigue mechanisms of baria-

silicate glass-ceramics under clinically relevant conditions.

The results show that the failure of materials with a low toughness such as baria-

silicate glass (0.7 MPa.ml12) and glass-ceramic with an aspect ratio of 3/1 (1.3 MPa.m12)

initiated from a cone crack developed during cyclic loading for 103 to 105 cycles. The

mean strength values of baria-silicate glass and glass-ceramic with an aspect ratio of 3/1

decreased significantly as a result of the presence of a cone crack. Failure of baria-

silicate glass-ceramics with an aspect ratio of 8/1 (K, = 2.1 MPa.m1/2) was initiated from

surface flaws caused by either polishing or cyclic loading. The gradual decrease of

fracture stress was observed in specimens with an aspect ratio of 8/1 after loading in air

for 103 to 105 cycles. A reduction of approximately 50% in fracture stress levels was

found for specimens with an aspect ratio of 8/1 after loading for 105 cycles in deionized

water. The mechanisms for cyclic fatigue crack propagation in baria-silicate glass-

ceramics are similar to those observed under quasi-static loading conditions. An

intergranular fracture path was observed in glass-ceramics with an aspect ratio of 3/1.

For an aspect ratio of 8/1, a transgranular fracture mode was dominant.














CHAPTER 1
INTRODUCTION

Ceramic materials are potentially useful for a wide variety of medical and dental

applications. They have several advantages over metals including their esthetic potential,

chemical inertness, low thermal conductivity, low thermal diffusivity, excellent

biocompatibility and their ability to be shaped using a variety of forming techniques.

However, their major drawback is their low fracture resistance. For dental prostheses,

relatively high clinical failure rates have been reported for several dental ceramics (1-3)

and this failure potential limits their use for dental applications. During mastication,

dental restorations are subjected to repeated loading, and crack initiation can occur after a

significant number of stress cycles that can lead to fatigue failure. In an attempt to gain a

better understanding of the failure mechanisms, previous investigators have focused on

the characterization of fatigue damage in several dental ceramics (4-10).

Hertzian indentation loading has been used in several studies to characterize the

mechanical properties of ceramic materials (9-12). Contact indentation loading is a

simple method for inducing surface damage and straightforward analytic relations can be

derived for the critical loads in terms of basic material properties (i.e., elastic modulus,

hardness, and fracture toughness) and contact radius (13). Clinically relevant parameters

can be controlled by the use of a repeated indentation loading test to simulate occlusal

intra-oral loads. However, the validity of the results for dental ceramics from the

Hertzian indentation test may be questionable when the loading conditions for the test do

not adequately reflect the loading characteristics that occur under situations in which









clinical prostheses have failed. Based on previous studies, fracture initiation occurred at

different sites depending upon the Hertzian loading conditions. The results of two

clinical studies indicated that crack initiation of clinically failed fixed prostheses occurred

either at the cement/ceramic interface or at the interface between the core and veneering

ceramics (14, 15). In contrast, failure of specimens from a blunt indentation test occurs

as a ring/cone crack or as a result of contact damage. The magnitude of load to failure

values determined from the blunt indentation tests appear to be too high compared with

normal occlusal forces. However, with a well-controlled testing protocol, valuable

information about fatigue failure parameters and mechanisms can be obtained from this

type of loading relative to potential dental applications.

Cyclic fatigue tests on baria-silicate glass and glass-ceramics using a spherical

indenter were selected to demonstrate the effect of crystal morphology on the mechanical

response to contact fatigue damage. The baria-silicate glass-ceramic system was chosen

for this study because (1) the system has a negligible difference in thermal expansion

between the glass phase and crystal phase, thereby sustaining little residual stress upon

cooling (16); (2) the baria-silicate glass phase and crystal phase have a nearly identical

composition; (3) the baria-silicate system can be nucleated homogeneously without the

addition of a nucleating agent; and (4) in higher crystal volume fractions, microstructures

with different aspect ratios can be produced (17).

The objectives of this study were to (1) test the hypothesis that after a specific

number of cycles, the mean strength of baria-silicate glass-ceramics subjected to cyclic

loading decreases because of critical flaw propagation; (2) test the hypothesis that the

crack growth rate of baria-silicate glass-ceramics under cyclic loading and low stress









conditions is strongly dependent on the number of stress cycles; (3) characterize the

influence of crystal morphology on fracture behavior as a function of number of loading

cycles; and (4) test the hypothesis that the crack growth rate of baria-silicate glass-

ceramics subjected to cyclic fatigue can be accelerated by environmental effects.














CHAPTER 2
LITERATURE REVIEW

Recent studies (9-12) have demonstrated clearly that fatigue damage occurs under

cyclic loading conditions in some commercial ceramics, and the strength of ceramics can

be compromised by slow crack growth of a critical flaw during repeated loading. When

the crack grows, the crack length increases over time until it reaches the critical value that

can cause a catastrophic fracture. Therefore, stress values well below the levels that

cause fast fracture can produce premature fracture of the ceramic structure. When the

crack grows under cyclic loading conditions, the crack growth rate can be approximated

using a simple power law equation that is dependent on the stress intensity ranges (18).

To select ceramics for specific structural applications, one should be aware of the crack

growth rates and cyclic fatigue behavior of ceramic materials during their clinical uses.

Damage induced by cyclic loading in brittle solids can lead to much higher crack growth

rates than those associated with environmental cracking under sustained loading

conditions (6). As a consequence, the service lifetime of brittle solid structures can be

reduced under repeated loading conditions.

Mechanisms of Fatigue Crack Growth in Ceramics

The crack growth behavior in ceramics is different than that in ductile materials.

Dislocation slip, which is a prominent crack propagation mechanism in metals, generally

does not occur in brittle solids at room temperature because of their covalent or ionic

bonds (19-21). Subcritical crack growth in brittle solids under cyclic loading involves

complex phenomena. In many brittle ceramics, there are no known differences between









the mechanisms of static and cyclic crack growth at low temperatures. There are some

general effects of cyclic loading on the growth of cracks, especially in ceramic materials

containing long cracks, i.e., compared with their microstructures. The pronounced

increase in crack growth rate caused by cyclic loading, compared with static test

conditions, is primarily caused by the breakdown of the bridging zone in the wake of the

crack tip. Residual stress produced by a thermal expansion mismatch between two

phases and elastic anisotropy between the crystals can promote the formation of

microcracks during the heating process and subsequent mechanical loading (18). The

crack growth rate can be related to the stress intensity factor using a power law

relationship:

da
-- = C(AK)m (2.1)
dN

da
log- = log C + m[log(AK)] (2.1a)
dN

where da/dN is the crack growth rate per cycle, AK is the stress intensity factor range

(Kmax-Kmin), and C and m are experimentally determined constants (18).

The typical crack growth rate behavior of materials is represented by a sigmoidal

curve on a logarithmic graph of da/dN versus AK, and this curve can be divided into three

distinct regions as shown in Fig. 2.1. The graph is linear in region II, which is consistent

with equation 2.1.

In previous studies, several crack growth mechanisms have been identified in

ceramics that relate to intrinsic microstructural damage and extrinsic crack-tip shielding

mechanisms (19-21). In metals, intrinsic mechanisms such as dislocation movement and

crack-tip plasticity are dominant, and they involve blunting and resharpening of the









crack. Conversely, extrinsic mechanisms result from formation of an inelastic zone

surrounding the crack wake, or from physical contact with the crack surfaces.



10-2 K
K4ic




10-4


Log AK


Figure 2.1. Schematic illustration of the fatigue crack propagation stages.

Mechanisms of Crack-Tip Shielding

Although shielding under monotonic loading results in a reduction in the local

driving force, the effect under cyclic loading may be different because the magnitude of

the principal driving force for fatigue crack progression is the difference between the

applied maximum and minimum stress intensity factors, AK. The difference in stress

intensity factor can be increased or decreased by changing Kmax and Kmin. While several

mechanisms of cyclic fatigue crack growth in ceramics have been proposed, Ritchie et al.

(20, 21) described the nature of ceramic fatigue according to two classes, intrinsic and









extrinsic mechanisms. Summarized in Table 2.1 are the possible mechanisms for crack

propagation under cyclic fatigue conditions in ceramics.

Table 2.1. Possible mechanisms of cyclic fatigue crack propagation in ceramics.


Mechanisms of cyclic fatigue crack growth in ceramics


(1) Extrinsic mechanisms
1.1 Degradation of transformation
toughening
1.2 Damage of bridging zone
Friction and wear of
unbroken
ligaments or whisker/fiber
reinforcements
Crushing of asperities and
interlocking zones
1.3 Fatigue of ductile reinforcing
phases


(2) Intrinsic mechanisms
2.1 Accumulated localized
microplasticity/microcracking
2.2 Mode II and III crack propagation
on unloading
2.3 Crack tip blunting/resharpening
2.4 Relaxation of residual stress


Source (20)

The shielding mechanisms involve crack deflection, zone shielding, contact

shielding, and a combination of zone and contact shielding. These mechanisms are the

principal means of increasing crack resistance in ceramics by reducing the local stress

intensity range, AK, which is a driving force for fatigue. Intrinsic mechanisms involve the

creation of a fatigue-damaged microstructure ahead of the crack tip, while the extrinsic

pathways involve a reduction of stress intensity via crack-tip shielding. Lathabi et al.

(22) and Dauskardt (23) indicated that the frictional wear mechanism accounts for cyclic

fatigue degradation in polycrystalline ceramics that exhibit grain bridging as a









toughening mechanism. This model is based on the sliding wear degradation of frictional

grain bridges, which reduces the toughening capacity from bridging under cyclic loading.

The accumulation of debris from wear processes was observed at the frictional sliding

sites during fatigue loading (24). It is suggested that this mechanism may be widespread

in polycrystalline ceramics with a crack-bridging toughening mechanism.

Brittle materials are known to fracture in monotonic compressive loading in two

different patterns, extrinsic and intrinsic modes (25). In an extrinsic mode, cracks extend

across planes of local maximum tension and propagate along a direction parallel to the

maximum plane of compression. The intrinsic mode occurs by a microscopic process

caused from an accumulation of pre-existing microcracks or weak interfaces and results

in a shear fault when the critical loading conditions are met. Ewart and Suresh (26)

confirmed the existence of crack growth in ceramics under cyclic compressive loading.

They used single edge-notched polycrystalline alumina specimens subjected to a

compressive cyclic load (26). Compressive stresses did not induce any crack growth,

since only tensile stress can cause crack propagation. They found that the formation of

particle debris within the crack was a characteristic feature of crack advance under cyclic

compression. The particles were generated by fracture along the grain boundaries and

their presence leads to crack closure effects. The authors reported that the crack growth

rate increased in alumina specimens when they periodically cleaned the debris particles

from between the crack faces. Therefore, their experimental results appear to be in

conflict with previous data (19-21), which suggested that the formation of debris particles

between the crack interfaces usually leads to a crack-tip opening condition or a wedging

effect. However, the cleaning process may be responsible to an increase in crack growth









rate by enhancing the stress corrosion at the crack tip. As a result, the increase in crack

growth rate in this study can be accelerated by either a wedging effect or stress corrosion.

The primary mechanism of cyclic fracture in alumina involves intergranular fracture,

unlike quasi-static fracture in which a transgranular fracture mode is prevalent. Grain

boundary microcracks nucleated during the compressive loading cycle within the process

zone at the notch tip lead to a reduction in the elastic modulus of the material in that

zone. The difference in the compliance of the material within the process zone and that

of the surrounding area can induce residual tensile stress during unloading of the

maximum compressive stress over a distance comparable to the size of the process zone.

Ewart and Suresh (26) hypothesized that residual tensile stress contributes to the growth

of fatigue cracks associated with grain boundary failure (18).

Contact Fatigue

For Hertzian contact of a sphere indenting a planar surface (18), one can calculate

the contact radius (a) and the pressure (p) within the contact area following these

assumptions: (1) the contact between the surfaces is nonconforming and frictionless; and

(2) the dimensions of the two bodies in contact are significantly larger than the contact

area.


a = (2.2)
4E*)


p (1- 2 )1/2 (2.3)
27ra a

1 1-v2 2
_1 2 (2.4)
E* E1 E2









where El and E2 are the elastic moduli, vi and V2 are the Poisson's ratio values of two

bodies, and E* is the effective elastic modulus.

The following general results are obtained from a Hertzian analysis (18, 27):

(1) The maximum normal tensile stress occurs on the surface at the contact perimeter,

and m = 1 (1 2v)p0, where po = P/Tra2
2

(2) The maximum shear stress (Tm) exists beneath the surface along the contact axis at
a depth of 0.48a,

Tm = 0.48po

(3) These results imply that failure should initiate near the contact perimeter at the
surface, while plastic yielding, crushing or microcracking is likely to occur beneath
the surface.



P





r




2R11


Figure 2.2. Hertzian contact loading.

Theoretically, when a hard spherical indenter is loaded normally on a flat, thick

elastic specimen, it produces a classical Hertzian cone crack when the applied stress is

greater than the critical value. The cone crack grows steadily with increasing load along

the path that is normal to the tensile stresses. Roesler (28) established the following

relation between the applied load P and cone base radius R:









P
= constant (2.4)
R 3/2

The distinctive characteristic of a blunt indenter is an elastic contact zone during

the time when the contact pressure increases monotonically with an expanding contact

circle. The contact becomes inelastic only if the contact pressure exceeds some critical

level for irreversible deformation prior to development of the cone-crack fracture (29).

The microstructure of a material significantly affects the damage response. Elastic-to-

plastic deformation occurs by microstructural coarsening of the crystal phase (9-12). A

zone of residual stress can also be generated once permanent strain is induced during

unloading (18). Residual stress can play a significant role in influencing the fatigue

responses as well as failure processes.

Fatigue can be considered either in terms of the accumulation of distributed damage

or as the propagation of a single crack. Lawn et al (24) identified the damage model

based on microcrack extension from the closed shear faults in the constraining

compressive fields. These faults account for a quasi-plastic deformation at the site of

loading that can lead to damage accumulation, material removal and wear, and strength

degradation in ceramic materials. The onset of quasi-plastic deformation can be

determined by the deviation from the nonlinear stress-strain curve (11, 12). Bonded-

interface specimens were used to determine the severity of subsurface damage caused

from a single indentation and contact fatigue. At the microstructural level, Lawn et al

(27) identified the character of shear faults as (a) cracking at the weak interfaces between

the glass phase and the second phase, (b) transgranular twinning, or (c) block slip in

monolithic structures. The general feature of these faults is the discreteness, localized by

the grain structure. The secondary microcracks are initiated at the ends of these faults.









Strength Degradation from Contact Damage

When blunt indentation produces the dominant flaws, the specimens will fail from

the indentation sites. Examination of fractographic features from previous studies (27,

30) confirmed that failure of fine grain and medium grain ceramics originated from the

cone base, and from within the subsurface damage zone in coarse-grain material. For

brittle materials, the presence of a cone crack will cause a rapid reduction in critical stress

for fast fracture of ceramic specimens. Strength degradation can be determined from the

classical Irwin-Griffith equation by using crack size calculated from the radius of a ring

crack, and a cone base angle. The critical stress intensity factor for the cone crack system

is identified as (29)

K =- = (2.5)
c2

where c is the cone crack length, and X is a dimensionless constant that depends only on

Poisson's ratio (29). The cone crack parameters are shown in Fig. 2.3. Although

equation 2.4 has the same form as that for sharp indentation, the significance of the

coefficient X is different. For the sharp indentation, there is a residual stress field to

accommodate the impression volume by expansion of the deformation zone against the

constraining elastic matrix. Therefore, X for sharp indentation is dependent on the nature

of the deformation, elastic modulus, hardness of the material, and indenter half-angle.

Environmental Effect

The fracture resistance of glasses and crystalline ceramics can decrease over time

under static loading and an active corrosive environment. The reduction in strength is

associated with the slow crack growth of pre-existing flaws caused by a stress corrosion

process. In silica, a chemical model for the interaction of the environment with









mechanically strained bonds at the tip of a crack is proposed as a mechanism for slow

crack propagation (31, 32). The process is described as consisting of the following

events:

(1) formation of a hydrogen bond between the water molecule and Si-O-Si bonds at the
crack tip;

(2) interaction of the lone pair of electrons from the oxygen atom in water (Ow) and the
Si atom;

(3) formation of two new bonds between Ow and Si, and between the hydrogen and
oxygen atoms from the silica molecule; and

(4) rupture of the hydrogen bond with Ow to yield surface Si-O-H groups on fracture
surfaces. These Si-O-H groups will not reform without any external energy being
supplied.



P


C I


R o /


Figure 2.3. Cone crack parameters.

There are three regions of crack propagation in a graph of velocity (V) versus stress

intensity factor (Ki). In region I, the crack propagation rate is sensitive to K1. The

sensitivity becomes moderate in region II and the crack growth is limited by the rate of

water transport to the crack tip. The crack propagation rate is independent of the

environment in region III. Subcritical crack growth can also occur in single crystal and









polycrystalline ceramics (33-36). In alumina and zirconia, slow crack growth caused by

water molecules has also been suggested (33-35). However, results from some studies

suggest another approach to explain how the active environment affects the crack

propagation rate (37-38).



Si Si


H. HH






















Besides static fatigue, failure that occurs by the simultaneous action of a cyclic

stress and chemical attack is termed "stress corrosion fatigue." In metals, mechanisms of
wH,. ,, H




Si Si t





Figure 2.4 Schematic illustration of the reaction of water and a strained Si-O-Si bond













stress and chemical attack is termed "stress corrosion fatigue." In metals, mechanisms of

environmental-assisted cracking are well documented (18, 40). Under cyclic loading, the

embrittling environment can accelerate the propagation of a flaw to a certain critical size,

and the crack propagation rate is enhanced as a result of exposure to a corrosive









environment. The mechanisms of fatigue crack initiation and propagation involve

accumulation of dislocations and electrochemical attack at the defective area.

For ceramic materials, the detrimental effect of cyclic loading in a wet

environment is well known (31-35). These results suggest that stress corrosion has a role

in controlling overall crack propagation because of the faster growth rates in water than

in air. The cyclic fatigue lifetime was shorter for alumina specimens tested in air

compared with those tested in vacuum (41). This result indicates that corrosion fatigue

can accelerate degradation associated with mechanical fatigue. In contrast, the results for

Si3N4 showed that the crack propagation mechanism under cyclic loading was

independent of any stress corrosion mechanism (42). However, the results from another

study indicated that the fatigue crack growth rate in Si3N4 can be enhanced by a corrosive

environment (43).

Fatigue Behavior of Dental Glass-Ceramics

Ceramics are routinely used for dental prostheses because of their superior esthetic

qualities and biocompatibility. Because of the relatively low tensile and shear strength of

dental porcelains, they are usually bonded to metal substructures made to fit prepared

teeth to minimize the risk of fracturing the core or veneering ceramics. However,

porcelain failures still occur and this represents a significant clinical problem for dental

practices when ceramic restorations are needed (1, 2). Chipping or cracking of porcelain

veneers has been reported to occur within 5 years after insertion (1, 2). Fracture is also a

major drawback of ceramic systems that have been developed to improve the optical

properties of ceramic prostheses. Relatively high clinical failure rates have been reported,

ranging from 35% after three years to 64% after four years for tetrasilicic-fluormica-

based glass-ceramic crowns cemented to molar teeth (3).











CRACK PROPAGATION











0 0 0 0





E 1D C B A


Region A: Elastically distorted bonds

Region B: Inelastic strain at the crack tip

Region C: Ruptured bonds with secondary bonds

Region D: Ruptured bonds without secondary bonds

Region E: Bonds in equilibrium with the environment

Figure 2.5. Schematic illustration of crack propagation in the presence of a chemical
wedge (38).

During normal mastication, dental prostheses are subjected to repeated loading

more than a thousand times per day. At stress levels typically well below the ultimate

stress of dental ceramics, unexpected fracture can occur after a significant number of

cycles sufficient to cause fatigue failure of the restorations. Because of the brittle

behavior of ceramic materials, they have little or no capacity to deform plastically under

high loads. In the past, fatigue behavior was not expected to be an important factor in the









control of failures. Recently, studies have demonstrated that dental ceramics can be

gradually degraded under cyclic loading conditions (4-7). By using a sharp indenter,

White (4) studied crack growth in feldspathic dental porcelain during repeated

mechanical loading. He reported that fatigue behavior of this porcelain was consistent

with the Paris model for cyclic mechanically induced crack growth (44). However, he

suggested that this might not be an appropriate model for dental situations because of the

high stress levels, which represents extreme conditions that occur rarely in vivo.

Several studies used a blunt indentation technique to investigate the cyclic fatigue

behavior of ceramics (5-7, 9). White et al. (5) used this technique in their study and they

determined the critical stresses necessary to produce elastic-plastic deformation and crack

initiation in feldspathic dental porcelain. They also confirmed the existence of fatigue

damage by performing a cyclic loading test (5, 6). They concluded that cyclic

mechanical fatigue caused irreversible damage to feldspathic porcelain. The severity of

accumulated damage was quantified with respect to the gradual strength degradation after

cyclic testing.

It is well known that the presence of water will cause strength degradation of

ceramic materials (8). White et al. (6) also investigated the effect of water on mechanical

fatigue in feldspathic dental porcelain and reported results for ambient, wet, and dry

environments. They reported that both static and cyclic mechanical fatigue significantly

reduced specimen strength, but these two test conditions did not affect each other.

Considerably greater damage and reduction in strength were noted for specimens tested

in water.









Besides feldspathic porcelain, other dental ceramic systems have also been

investigated (7). Three ceramics, glass-infiltrated alumina (In-Ceram), a leucite-based

glass-ceramic (Optimal Pressable Ceramic), and a leucite-based glass-ceramic (IPS

Empress), were subjected to cyclic loading and then fractured in wet and dry

environments. The decrease in fracture stress caused by cyclic loading was in agreement

with the results of a previous study (6). This effect was more pronounced for specimens

exposed to an aqueous environment.

Mechanism of Fatigue Damage

The existence of fatigue damage in ceramics raises considerable concern regarding

the life prediction of brittle materials. It has been shown that subcritical crack growth

occurs under cyclic loading conditions with stress levels well below the critical value for

catastrophic fracture. The nature of crack propagation and crack growth resistance of

these brittle materials has been widely investigated. The results of these studies suggest

that several factors and mechanisms are involved in the fatigue behavior of different

types of ceramic materials (45, 46).

For dental ceramics, Peterson et al. used bonded-interface specimens to evaluate

the role of microstructure and related failure mechanisms of specimens subjected to

repeated loading (9, 10). A series of micaceous glass-ceramics were prepared by

controlled heat treatment with mica crystal diameters between 1 and 10 |tm and aspect

ratios between 3/1 and 9/1. The polished surfaces of two ceramic block specimens were

bonded together with cyanoacrylate adhesive. After cyclic loading, two damage modes

were observed, tensile-induced cone cracking in a fine-grain material and quasi-plastic

deformation in a coarser grain glass-ceramic. They concluded that microstructure is a









controlling factor in determining the nature and degree of damage accumulation in dental

ceramics. They also used other dental glass-ceramics in their study, such as Vita In-

Ceram, a glass-infiltrated alumina, Vita Mark II, a modified feldspathic porcelain, and

yttria-stabilized tetragonal zirconia. They found that a quasi-plastic response was

observed in zirconia and glass-infiltrated alumina. The most brittle response (cone crack

formation) was observed in modified feldspathic porcelain.

The bonded-interface technique has been used in earlier studies (11, 12) in which

the damage beneath a Hertzian contact field was examined in commercial mica-

containing glass-ceramics compared with the glass specimens. For the glass samples,

fatigue failure was caused by slow growth of cone cracks. On the other hand, the fatigue

process in glass-ceramics was enhanced by formation of microcracks developed from

shear stress along the weak interfaces between the mica platelets and the glass matrix.

The effect of environment on the fracture behavior of dental ceramics has also

been studied (47-50). Specimens made from aluminous and feldspathic porcelain were

tested under three-point loading (49). The authors found that specimens that were tested

under dry conditions were approximately 27% stronger than those that were tested in

distilled water.

Previous studies showed that dental ceramics could be weakened by fatigue

damage (4-7). The investigators attempted, with little success, to identify the

mechanisms responsible for the degradation of the materials during cyclic fatigue. They

also used a tungsten carbide indenter to produce contact damage within the specimens.

Tungsten carbide has a remarkably higher elastic modulus than most other materials that

are used to restore missing tooth structure. Thus, the use of a tungsten carbide indenter









can be considered as an extreme condition and the experimental results might not be

representative of the oral environment.

Several previous investigators analyzed the fracture failure of dental ceramics in a

brittle mode because they thought that ceramic materials could not deform plastically.

Therefore, only a limited number of investigators have analyzed the fatigue damage

caused from repeated loading conditions. The lack of information in this area is obvious.

In addition, all previous data has been obtained on materials that have thermal expansion

anisotropy, which leads to development of a residual stress upon cooling. The residual

stress may affect the damage modes observed. Thus, a study that can minimize

extraneous effects from microstructural property is preferred.

Specific Aims

The specific aims of this study were to

(1) test the hypothesis that, at a specific number of cycles, the mean flexural strength
of baria-silicate glass-ceramics subjected to cyclic loading is rapidly decreased
because of critical flaw propagation;

(2) identify the controlling mechanism of crack propagation in baria-silicate glass-
ceramics based on two crystal aspect ratios;

(3) test the hypothesis that the crack growth rate of baria-silicate glass-ceramics under
cyclic loading and low stress conditions is strongly dependent on the number of
stress cycles;

(4) characterize the influence of crystal morphology on fatigue failure as a function of
number of loading cycles; and

(5) test the hypothesis that the crack growth rate of baria-silicate glass-ceramics
subjected to cyclic fatigue can be accelerated by environmental effects.














CHAPTER 3
MATERIALS AND METHODS

Preparation of Baria-Silicate Glass-Ceramic Specimens

Baria-silicate glass plates containing 39.5 wt% BaO and 60.5 wt% Si02 were

obtained from Coming Inc. (Coming, NY). Bars, approximately 25 mm long, 5.1 mm

wide, and 2.5 mm thick, were cut from these plates using a diamond cutting saw

(IsometTM, Buehler, IL).

Heat Treatment Schedules

The glass bars were heat-treated to produce glass-ceramic specimens with desired

microstructures. One hundred and forty bars were nucleated and crystallized to produce

two different microstructures. The treatment times and temperatures are listed in Table

3.1. Seventy glass specimens that were not heat treated were used as control specimens.

Table 3.1. Heat treatment schedules.

Nucleation Crystallization
Aspect ratio
Temp (C) Time (h) Temp (C) Time (h)

700 0.75 825 1.5 3.6

700 1.0 825/1050 24/24 8.1


Microstructural Analysis

The percent crystallization of baria-silicate glass-ceramics was determined using

the point-counting technique. Specimens were etched in 1% HF for 10 s, rinsed in

deionized water, dried and coated with Au-Pd for scanning electron microscopy. Sixteen









images for each aspect ratio group were recorded [4 specimens x 2 cuts (900 and 450) x 2

surface areas]. The aspect ratios were determined from the major and minor dimension

of discernible crystals in SEM micrographs. All micrographs were produced at 4000X.

Surface Preparation

Bar-shaped specimens, 25 mm long, 5 mm wide, and 2 mm thick, were prepared

from baria-silicate glass-ceramics. The glass and glass-ceramics bars were ground and

polished through 600-grit silicon carbide abrasive using a hand lapper (South Bay

Technologies, San Clemente, CA). The surfaces to be subjected to cyclic fatigue loading

were polished through 1200-grit silicon carbide and 1 |tm alumina paste. The specimen

edges were rounded using 1200-grit silicon carbide to reduce the risk of premature failure

at stress concentration sites. All specimens were annealed at 6000 C for 2 h after

polishing, and ultrasonically cleaned in ethanol for 5 min before testing.

Density

The density of each material was determined from the mass to volume ratio. Each

specimen was dried in furnace and its dry weight was measured on a precision balance

(Mettler H31, Mettler-Toledo Inc., NJ). Three volumetric readings were made for each

specimen to determine the average volume using a pycnometer (Model 1330,

Micrometrics, Norcross, GA). The density of each specimen was then calculated.

Young's Modulus and Poisson's Ratio

Groups of four specimens each were prepared for each aspect ratio and used to

determine elastic modulus and Poisson's ratio values using an ultrasonic device (Nuson

Inc., Boalsburg, PA). The transducers were attached to the specimens to generate the









longitudinal and shear waves. Elastic modulus and Poisson's ratio were calculated from

the shear and longitudinal velocity values.

Test Methods

Mechanical Fatigue Test

A cyclic load was applied to baria-silicate glass-ceramic specimens using a servo-

hydraulic testing machine (Model 1350, Instron Corp., Canton, MA). A schematic

illustration of the cyclic fatigue fixture is shown in Fig. 3.1. In an attempt to simulate test

conditions that approximate oral conditions, a low loading frequency (&3 Hz) and a low

load level (_< 200 N) were used. A blunt Type 302 stainless steel indenter was used to

apply the load on the specimen surface. The diameter of the indenter used in this study

was 4.76 mm, which is approximately equal to the cuspal radii of molar and premolar

teeth. Ten bar specimens were subjected to one of the following numbers of loading

cycles: 0, 103, 104, and 105 cycles. After cyclic loading, specimens were observed for

surface damage using an optical microscope, and then they were subjected to four-point

bending and observed for subsurface damage.

Strength Degradation Test

A four-point bending test was used to quantify the severity of the strength reduction

caused by the repeated loading test. Glass-ceramic specimens were fractured using a

universal testing machine (Instron Model 1125) at a crosshead speed of 2.5 mm/min. A

four-point flexure apparatus with an upper span of 6.67 mm and a lower span of 20 mm

was used. The load to failure, P, was used to calculate the fracture stress, of, using the

following equation:

3Pa
3Pa (3.1)
Sbt2









where P is the failure load, a is the horizontal distance between support and loading

points, b is the bar width, and t is the bar thickness.



P: 0-200 N







Loading cylinder

Steel ball


Steel support



Figure 3.1. Schematic illustration of cyclic fatigue fixture.

The environments used for the cyclic fatigue test were ambient room temperature

(250C) and deionized water at 37C.

Fractographic Analysis

The critical flaw size was measured using an optical microscope. The fracture

toughness of each material was determined from specimens in the control group (0 cycle)

using the Griffith-Irwin equation (51):

Kc = YC(c)1/2 (3.2)

where Y is a geometric constant, cy is the calculated fracture stress, and c is the measured

crack size within the fracture surfaces. The crack depth was calculated as the square root

of the depth times the half width. A schematic illustration of fractographic features is









shown in Fig. 3.2 (52). Surface and subsurface damage of specimens subjected to cyclic

loading were identified using an optical microscope and SEM images.

Statistical Analysis

An analysis of variance was performed to determine if a statistically significant

difference existed between mean values of the independent variables. The null

hypothesis (Ho) was that there was no statistically significant difference between mean

values. The alternative hypothesis (HA) was that not all means were equal. The

significance level (uc) was set at 0.05. P-values were calculated from the data. If the p-

value was lower than the significance level (c = 0.05), the null hypothesis was rejected,

and a multiple comparison analysis was performed to determine the order of the

statistical subsets.




Hackle region
Mist region c = (ab)1/2
Mirror region
Critical crack ." ::. ..









2bc-l
--- .. -------"---




rcb r'cb



Figure 3.2. Schematic representation of fracture surface features: ai and bi are original
crack depth and width; and acr and bcr are critical crack depth and width.














CHAPTER 4
RESULTS AND DISCUSSION

Physical properties of baria-silicate glass and glass-ceramics are summarized in

Table 4.1. Young's modulus increased from 64.8 GPa for the glass to 80.5 GPa for the

glass-ceramic with a high aspect ratio. These values are in good agreement with those

reported by Freiman et al. (17).

Table 4.1. Physical properties of baria-silicate glass-ceramics.


Materials Aspect Density Young's Poisson's Crystal
Ratio (g/cm3) Modulus (GPa) Ratio Volume Fraction

Glass 3.99 0.01 64.8 0.29 0%

Glass-ceramic 3.6 3.93 + 0.02 75.3 0.28 75%
(AR 3)

Glass-ceramic 8.1 4.02 0.01 80.5 0.27 73%
(AR 8)



The highly crystalline (> 70 vol%) glass-ceramics were produced with two aspect

ratios (AR) using heat treatments specified in Table 3.1. For glass-ceramic materials, the

crystal diameter ranged from 1.5 to 11 ptm for the AR3 group, and from 0.5 to 15 ptm for

the AR 8 group. The microstructures of the two glass-ceramics are shown in Fig. 4.1.

The crystals in AR 3 group specimens appear spherulitic with an aspect ratio of 3.6:1.

Crystals in the AR 8 group specimens are elongated with an approximate length to width

aspect ratio of 8:1.

















(a) AR 3















(b) AR 8








SEM images of baria-silicate glass-ceramic microstructures with aspect
ratios of 3/1 (a) and 8/1 (b) after etching with 1% HF for 10 s (4000X).


Figure 4.1.










Table 4.2. Mean flexural strength and standard deviation for baria-silicate glass-
ceramics.


Number of Environment Glass AR 3 AR 8
cycles (MPa) (MPa) (MPa)

0 Air 96 14 (A) 150 20 (A') 230 14 (A")

103 Air 61 20 (B) 138 20 (A') 225 12 (A")

104 Air 63 19 (B) 108 33 (B') 208 20 (B")

105 Air 51 + 14 (B) 80 18 (C') 209 17 (B")

0 Deionized water 98 + 17 (a) 150 + 20 (a') 185 + 7 (a")

103 Deionized water 53 13 (b) 118 24 (b') 201 + 19 (a")

104 Deionized water 45 11 (b) 100 14 (c') 171 24 (b")

105 Deionized water 50 + 9 (b) 82 15 (d') 110 15 (c")

() The mean values in a column with the same letter are not significantly different












300

250 -

S200 AR 8 .

I 150 AR3 -- A.....
....................... l ....T
AR3- W ,... | ...W
S Glass
00 G l"..................... ... .. ...
.. Glass A
50 ..................... .....................

0 1
0 1000 10000 100000
Loading cycles


Figure 4.2. Mean strength of baria-silicate glass-ceramics as a function of number of
loading cycles (- Tested in air; .****** Tested in water).

Strength Degradation

The mean flexural strength and standard deviation for baria-silicate glass and glass-

ceramic specimens that were cyclically loaded at ambient room temperature and in

deionized water are summarized in Table 4.2. The strength variation as a function of

number of loading cycles is shown in Fig. 4.2.

Based on Duncan's multiple comparison tests, the mean strength of the group with

an aspect ratio of 8/1 (AR 8) was significantly greater than that of each of the other

groups at each number of cycles in air (p < 0.05). The mean flexural strength of group

AR 3 was also significantly greater than the mean of the glass group at each number of

cycles in air (p < 0.05). The mean flexural strength of glass specimens decreased

significantly after 103 cycles (p < 0.05). For glass-ceramic specimens with the lower









aspect ratio (AR 3), the mean strength decreased after loading for 104 cycles (p < 0.05).

A reduction in mean strength of group AR 8 specimens was observed after 104 and 105

cycles.

The decrease in strength of baria-silicate glass and AR3 specimens that were

cyclically loaded in deionized water shows a similar trend compared with those exhibited

by the specimens tested at room temperature, but the decrease in strength occurred at

fewer cycles. For the glass and AR 3 groups, the strength decreased after loading for 103

cycles. The decrease in mean strength of specimens with an aspect ratio 8/1 was observed

at 103 cycles. The mean strength of group AR 8 specimens tested in air did not decrease

as much as the mean strength of glass and group AR 3 specimens. However, in deionized

water, the decrease in strength of group AR 8 was significant. After 105 cycles, a

decrease of approximately 50% in mean strength was observed for AR 8 group

specimens.

Fractographic Analysis

Baria-Silicate Glass

Fractographic analysis can be used as a characterization tool to detect the type of

strength-limiting cracks formed by cyclic loading. Based on fracture surface analysis of

cyclically loaded baria-silicate glass specimens, four crack configurations were observed

and a schematic illustration of two of these failure origins is shown in Fig. 4.3. The first

type (I) was a concentric ring and cone crack type (Fig. 4.4), which initiated from the

base of the cone crack and propagated approximately perpendicular to the principal

direction of tensile stress (point B in Fig. 4.3). The second type (II) was a triangular-

shaped crack that formed during four-point flexure at the base of the cone crack (point A-

A in Fig. 4.3) and subsequently propagated along the plane of a cone that developed









during cyclic loading (Fig. 4.5). The third type (III) was a semi-elliptical surface crack

caused by cyclic loading damage at the site of loading (Fig. 4.6). The fourth crack

configuration (IV) was a semi-elliptical crack caused by polishing (Fig. 4.7). The first

two types were observed in the glass and AR 3 specimens subjected to cyclic loads

ranging from 103 to 105 cycles. The third crack type was observed in group AR 3, AR 8

and glass specimens tested in deionized water for 104 and 105 cycles. The fourth crack

type was observed in specimens in the control group (0 cycle) and in specimens in which

the damage from cyclic loading was not as severe as the surface finishes.


B (Ta


(1) Triangular crack originated from point A-A and propagated to point A' on a
A'ABED plane to form a critical crack leading to fast failure
(2) Failure started at either of the two point B sites

Figure 4.3. Schematic illustrations of selected failure origins from cone cracks in glass
and glass-ceramic specimens (Top view).
























(a) 25X


(b) 80X

Figure 4.4. SEM images of a cone crack formed in deionized water after 104 loading
cycles. Fracture typically originated from either point B.


.111OF6.








L.-a.:lIiig sura. : [ e


Ril? c i'rack-
x


ara~la rc


(a) Loading and fracture surface images (100X)










(b) Fracture surface images (30X and 100X)
Figure 4.5. SEM images of a triangular crack in a glass specimen caused by cyclic
loading in air for 105 cycles. Fracture initiated from points A-A.






















(a) 25X (b) 100X


Figure 4.6. SEM images of a semi-elliptical surface crack in a glass specimen caused
by cyclic loading in deionized water for 105 cycles. The dashed line in (b)
outlines the region of the critical crack.


(a) 75X


Cicll



au


(b) 200X


Figure 4.7.


SEM images of a typical semi-elliptical crack in a glass specimen caused
by polishing.









Strength Degradation Associated with a Cone-Crack

Few studies have been reported on the analysis of strength degradation caused by a

Hertzian indentation crack (53,54). For well-developed cone cracks (Fig. 4.8),

dimensional analysis (29) indicates that:

P
R3/2 (v, 8)Kc (4.1)

where P is the indentation load, R is the radius of the cone base, and Ua (v, P3) is a

dimensionless constant dependent on Poisson's ratio (v) and the cone-crack angle (P). Kc

is the toughness of the material in term of the critical stress intensity factor.

Lawn et al. described the strength degradation (of) from cyclic loading related to

the cone base radius (Fig. 4.8) as (53):

Kc = C,7(xR)1/2 (4.2)


(3)R (= 2 (4.3)


where Q(P3) is a dimensionless constant dependent on P.

Evans (54) assessed the effect of conical cracks on the strength of ceramic

components (Fig. 4.9). He analyzed the stress necessary to extend a Hertzian crack,

which should be inversely proportional to the square root of the crack depth, L.

Kc = (L1/2) (4.4)
Z

-Z = U f (LI/2/KC) (4.5)
Y


r1/2, and Z is a flaw morphology parameter.


where L = G sin a, Y'













2a


Figure 4.8. Schematic illustration of a cone crack induced by a Hertzian contact load.



C ..---------> ------ --------- a



/\



Figure 4.9. Fracture origin associated with a Hertzian cone crack as suggested by
Evans (54). Failure initiates at point A.









An analysis of the stress intensity factor for a triangular crack inclined at different

angles to the tensile axis was reported by Murakami (55). The crack area projected in the

direction of the maximum tensile stress was used to calculate the stress intensity factor

along the crack front using the following equation:

KeO = Co_[7r(A,)12 ]1/2 (4.6)


C 1/2 K max (4.7)
S[(A p)12 ]1/2 (4.7)


where Ap is the crack area projected in a plane perpendicular to the direction of the

tensile stress and Co is a constant obtained from the above equation. Kemax is the critical

stress intensity factor in the polar coordinate system (r, 0).

Based on fracture surface analysis, the procedures used to measure the crack sizes

and calculate geometric constant values were described (equations 4.3 4.7).

Summarized in Table 4.3 are crack sizes and geometric constants calculated from the

known fracture toughness of baria-silicate glass (0.7 MPa*m1 2), and equations 4.3, 4.5

and 4.7 for the triangular crack (the second crack type). Critical crack sizes were also

calculated using the Griffith-Irwin equation (equation 3.2) and Y values are included in

Table 4.3 (51). The crack size as a function of loading cycles of glass specimens is

shown in Fig. 4.10.

Few studies have analyzed the potential methods to predict strength degradation

caused by cone cracks (53-56). Based on the cone crack configuration, failure in flexure

is expected to initiate from one of the two diametrically opposite positions on the base

rim of the cone (point C-C in Fig. 4.3) in the symmetric plane containing the axes of

indentation loading. The failure analysis for this condition was developed in terms of









fracture mechanics principles, which assume that the Hertzian cone crack parameter,

RQ(cL), is the critical flaw size (53). The approximation of strength degradation by the

cone crack as described in another study was also considered (54). Although the author

(54) agreed that the fracture should originate from a position at the base exposed to the

maximum stress intensity (point A in Fig. 4.9), the decreased strength values were

actually determined by using the depth of the cone as the critical crack size. Based on the

fracture surface analysis in the present study, we observed two crack configurations that

could not be clearly characterized by the methods described previously (53-54). For

example, the observed triangular crack that was likely formed during four-point flexure.

From the fracture surface markings, we observed two symmetric points on the cone base

as failure origins. Cracks propagated from these two points to join each other and form a

critical crack for catastrophic failure. Thus, failure of specimens in this study did not

start from the site (point C-C in Fig. 4.3) described in previous studies (53-54) and

resulted in formation of a different crack configuration.

Table 4.3. Crack sizes and geometric factors for glass calculated from equations 3.2 and
equations 4.3-4.7.


No. OR L (im) Ap (ab)1/2 Y'/Z Co71/2 Y
of (im) (tm) (tm)
cycles (Eq. 4.3) (Eq. 4.5) (Eq. 4.7) (Eq. 3.2) (Eq. 4.5) (Eq. 4.7) (Eq. 3.2)

103 54 120 109 113 1.18 1.24 1.22
(9) 12 +5 +10 +7 0.16 0.10 0.12

104 56 130 118 128 1.15 1.17 1.17
(8) 11 +16 + 15 +8 +0.10 +0.07 +0.11

10s 64 149 153 144 1.13 1.11 1.15
(9) 36 +17 + 13 +22 0.27 0.28 0.26
The mean critical crack size for the control group (0 cycle) is 35 14 [tm.
() Number of specimens in each group











(a) Air


1000


10000


Loading cycles




(b) Deionized water


0 1000 10000 100000


Figure 4.10.


Loading cycles


OR: Crack size for equation 4.3; L: Crack size for equation 4.5
Ap: Crack size for equation 4.7; (ab)1/2: Crack size for equation 3.2

Crack sizes as a function of cyclic loading for glass specimens: (a) Tested
in air; (b) Tested in deionized water. The legends in the graphs refer to
the parameters used in Eqs. 3.2, 4.3, 4.5, 4.7 and in Table 4.3.


150
N
. 100
u


100000


150


- 100


L
- A Ap
0 (ab)1/2



om a




1111111*









As a result, it is unrealistic to use the methods described in those studies to

characterize the triangular crack. Therefore, comparisons between crack size and the

geometric constant values determined from each method have been made to determine

the optimal technique for describing this crack system.

From Table 4.3, the mean crack sizes calculated from equations 3.2, and 4.7

[(ab)1/2, Ap] were comparable and not significantly different (p > 0.05). Y and Ct17/2

values were also comparable. The crack size calculated from Equation 4.3 was

significantly smaller than that calculated from the other equations. Lawn et al. (53)

estimated that Q(P3) values range from 0.25 to 0.30 for soda-lime silica glass. The Q(P3)

values of 0.11 for soda-lime glass and 0.34 for alumina were obtained from another study

(56). Because of different Q(P3) values reported for different materials, they suggested

that Q(P) is a material constant (56). In our study, the Q(P3) value was similar to that

reported by Lawn, but with a larger standard deviation (SD). The reason for the large SD

is most likely associated with the failure of glass specimens that does not initiate from

one of the two diametrically opposite positions (either point C in Fig. 4.3) on the base rim

of the cone coincident with the symmetry plane containing the axes of initial contact

loading (53). The second possible reason is that the curvature of the cone base must be

considered when the cone crack is large.

The Y'/Z values calculated from equation 4.5 are comparable to Y and C071/2

values. Evans reported that the Z value ranged from 1.3 to 1.5 for glass, alumina and

silicon carbide (54). He also suggested that the values obtained in his study were similar

to that for a surface crack of similar depth with a length to depth ratio of 3/1. For a









length to depth ratio of 3/1, the value of flaw parameter (Q) is 1.75 (with a/b = 0.3). The

Q value can be related to the critical stress intensity factor (40):


Kc = Yoa (4.8)


The Z value in equation 4.5 is reported to be a dimensionless constant that depends

on the configuration of the crack (54), i.e., a correction factor for a semi-elliptical crack.

However, the approach using this Z value and the depth of the critical crack seems to be

inappropriate because only one dimension of a cone crack is considered. For critical

crack size determination, width and depth are usually included to obtain an accurate crack

size value. The location of the point at which fracture initiates is also critical in

determining the crack size and a geometric constant value and should be identified during

failure analysis processes.

Summarized in Table 4.4 are the mean crack sizes and standard deviations

determined from equation 3.2 and equations 4.3-4.7 of baria-silicate glass specimens that

were cyclically loaded in deionized water. We observed the same crack patterns in the

groups subjected to 103 and 104 loading cycles. However, contact damage within the

surfaces of four specimens in the 104 cycle group and eight specimens in the 105 cycle

group was clearly evident and different from that observed in specimens tested in air.

Failure of these specimens tested in deionized water initiated from a semi-elliptical crack

(type III) caused by contact surface damage (Fig.4.6). The mean critical crack sizes of

these groups (104 cycle and 105 cycle group) were 194 55 and 135 32 |tm,

respectively.

From Table 4.4, the values of Y, Y'/Z, and Cot1/2 tend to decrease gradually with an

increase in the number of cycles from 104 cycles to 105 cycles. We hypothesize that the









depth of the crack tip is responsible for reduction in these three constant values, i.e., there

is a decrease in tensile stress, as the cracks grow deeper into the structure. The cone

crack in specimens tested in deionized water propagated to a greater depth beneath the

contact surface than that in specimens tested in air because of the stress corrosion effect.

Therefore, critical crack sizes of specimens tested in deionized water were larger and

resulted in a decrease in constant values. We recalculated the stress values according to

the depth of the critical cracks, local, using the following equation:

2x
cal = x ( -) (4.9)

where cf is the stress at fracture, b is the thickness of a specimen, and x is the depth of

the critical crack.

Table 4.4. Crack sizes and geometric factors for glass specimens tested in deionized
water.


No. QR L (im) Ap (ab)"12 Y/Z Co"1/2 Y
of (min) (tmin) (inm)
cycles (Eq. 4.3) (Eq. 4.5) (Eq. 4.7) (Eq. 3.2) (Eq.4.5) (Eq.4.7) (Eq.3.2)

103 54 138 105 104 1.12 1.27 1.27
(9) 17 +39 +9 +10 +0.17 0.21 0.21
104 71 163 149 147 1.14 1.21 1.21
(4) 29 +8 +11 +14 0.25 31 0.31
105 52 169 166 168 0.98 0.98 0.98
(2) 12 +24 +14 0.6 +0.18 +0.11 +0.11

The mean critical crack size for the control group (0 cycle in deionized water)
is 31 12 [tm.
() Number of specimens in each group









By applying equation 4.9 to equation 4.3 4.7, and 3.2, we obtain corrected values

for Y, Y'/ Z, and Co7 /2. In Table 4.5, Y increases from 0.98 0.1 to 1.1 0.1 and Cot1/2

increases from 0.98 0.1 to 1.1 0.2 for specimens loaded for 105 cycles when local is

substituted into equation 4.7 and 3.2. Although the "constant" values increase after

applying the corrected local into equation 4.3-4.8, the increase is still not significant. The

possible reason for the decrease of the geometric constants in this group of specimens

that contain the largest and deepest critical flaw size may be the transition from a surface

flaw to an internal flaw condition. For example, the Y value decreases from 1.26 for

surface flaws to 1.12 for internal flaws. The decrease in the Y value in this study

suggests the dominance of internal flaws. However, there were only two specimens in

this group that failed in this manner. Therefore, no conclusive explanation can be made

to accept or reject the proposed hypothesis that the depth of the crack is responsible for

reduction in the constant values.

Table 4.5. Geometric factors of glass specimens tested in deionized water with adjusted
stress.


No. n Y'/Z Co00/2 Y "corr Y'/Zeorr CoEI1/2 Ycorr
of (Eq.4.3) (Eq.4.5) (Eq.4.7) (Eq.3.2) (Eq.4.3) (Eq.4.5) (Eq.4.7) (Eq.3.2)
cycles

103 0.24 1.12 1.27 1.27 0.26 1.17 1.32 1.33
(9) 0.09 0.17 0.21 0.21 0.09 0.18 0.22 0.22
104 0.19 1.14 1.21 1.21 0.26 1.23 1.29 1.30
(4) 0.12 0.25 0.31 0.31 0.15 0.33 0.34 0.32
105 0.15 0.98 0.98 0.98 0.18 1.07 1.08 1.07
(2) 0.04 +0.18 +0.11 +0.11 +0.04 +0.19 +0.17 +0.12









Based on data in Tables 4.3 and 4.4, the constant values Q (0.19 0.24) and Z (1.51

- 1.66) appear to be in the range reported in previous studies (53, 54) with higher SD

values. However, the previous investigators did not clearly identify the failure origins of

the specimens and the strength degradation analyses were based only on the cone crack

geometry that can be determined from fracture mechanics equations. The shape of

critical cracks and failure origin sites were not identified. It is important to identify the

failure origins; otherwise, one cannot determine the correction factors for crack geometry

and free surfaces. Therefore, we can obtain only minimal information from the constant

values Q and Z. In contrast, the constant values C071/2 and Y were determined from the

surface area of critical cracks. An assumption was made for the shape of the critical

flaws to simplify the calculation methods. For a constant value of C0t1/2, a triangular

crack shape was assumed to calculate the real surface area of the critical crack that

propagated from point A-A to A' on a plane of a cone crack. The projected area in the

direction of the maximum tensile stress was used to determine the geometric constant

value for Mode I crack propagation. For the Y value, we measured the size of a

triangular crack and calculated the crack area directly from the fracture surface. The

constant values C071/2 and Y are in good agreement for glass specimens. The surface

area approach is more rational because the actual size and shape of critical flaws are

included in the calculation procedures. These constant values also represent the flaw

characteristics, compared with the regular semi-elliptical or semi-circular flaws.

However, as discussed earlier regarding the depth of critical cracks, we should consider

the surface-to-internal flaw characteristics as the cone crack propagates to a greater depth

beneath the contact surface either because of stress corrosion or high load conditions.









Baria-Silicate Glass-Ceramics

For baria-silicate glass-ceramics with an aspect ratio of 3/1, we observed the same

four patterns of crack configurations (Fig. 4.11 and 4.13) as those observed in glass

specimens, i.e., cone-cracks and surface damage cracks. Summarized in Table 4.6 are the

crack sizes and geometric factors calculated from equations 4.3-4.8 for group AR3

specimens tested in air. Crack sizes as a function of loading cycles for group AR 3

specimens tested in air and in deionized water are shown in Fig. 4.12. In this group, cone

cracks were observed in specimens subjected to cyclic loading in air for 104 and 105

cycles. No cone crack or contact-induced surface damage was observed in specimens

subjected to loading for 103 cycles. For specimens subjected to loading at 103 cycles, the

mean crack size was 57 + 19 |tm, which is not statistically significantly different ( p >

0.05 ) from that of the control group. For specimens subjected to cyclic loading at 104

and 105 cycles in air that failed from a cone crack, Y, Y'/Z, and Cot1/2 values are greater

than those calculated for the glass specimens tested in air and the crack sizes are smaller.

One possible reason for the greater constant values is that the cone cracks, which

developed in the AR 3 group, were not as deep as those found in the glass group and act

as surface cracks. If we consider the shape of the fracture-initiating flaw as a surface

crack, then recalculation of a new crack size results in a larger crack dimension. As a

result, the Y factor calculated from this new crack size using equation 4.8 is 1.26, and

represents the fracture mechanics correction factor for free surfaces (57). Another

possible reason is the influence of the local residual stress that develops during cyclic

fatigue testing, which one must take into account to calculate the Y value.









For group AR 3 specimens, the fracture toughness calculated from the control

group using fractographic analysis is 1.26 + 0.05 MPa.m1/2. The increase in fracture

toughness from 0.7 MPa.m1/2 in the base glass to 1.26 MPa.m1/2 in the AR 3 group is

responsible for the increase in the numbers of cycles needed to initiate the cone-crack

(from 103 cycles to 104 cycles). The fracture surfaces of group AR 3 specimens subjected

to cyclic loading between 103 to 105 cycles are shown in Fig. 4.11.

For group AR 3 specimens tested in deionized water, cone cracks first appeared in

specimens after 103 loading cycles, i.e., fewer cycles than required for specimens tested

in air. The geometric constant values are similar to those of specimens tested in air.

However, the mean critical crack sizes of specimens loaded for 104 and 105 cycles in

deionized water were significantly larger than those in specimens tested in air for the

same number of loading cycles. The geometric constant values decrease as a result of

increasing crack sizes at a constant stress level.

Representative SEM images of fracture surfaces of group AR3 specimens loaded in

deionized water are shown in Fig. 4.13. The contact-induced surface damage can be seen

on the loaded surfaces of specimens subjected to 104 cycles. The contact damage was

more severe in specimens subjected to 105 cycles. As shown in Fig. 4.13b, a surface

impression is noticeable, and the zone of contact damage beneath the loading site is

confined within the cone crack. The failure of this specimen did not start from the cone

crack, but instead initiated from within the surface contact damage area. More details

about the failure initiation process are discussed in the section on fatigue failure

mechanisms.









Table 4.6. Crack sizes and geometric factors of group AR 3 specimens tested in air.


No. OR L (im) Ap (im) (ab)1/2 Y'/Z Con1/2 Y
of (min) (inm)
cycles (Eq. 4.3) (Eq.4.5) (Eq. 4.7) (Eq. 3.2) (Eq. 4.5) (Eq. 4.7) (Eq. 3.2)
104 63 89 96 81 1.50 1.34 1.54
(6) 23 +29 +25 +21 0.14 0.06 0.11
105 90 112 123 96 1.53 1.28 1.53
(8) +53 +20 +17 +10 +0.31 +0.11 +0.11

The mean critical crack size for the control group (0 cycles) is 53 11 tim.
() Number of specimens in each group



Table 4.7. Crack sizes and geometric factors of group AR 3 specimens tested in
deionized water.


No. nR L (im) Ap (ab)1/2 Y'/Z Co71/2 Y
of (min) (min) (inm)
cycles (Eq. 4.3) (Eq. 4.5) (Eq. 4.7) (Eq. 3.2) (Eq. 4.5) (Eq. 4.7) (Eq. 3.2)

103 53 77 80 63 1.48 1.44 1.63
(4) 14 +19 +17 +17 0.20 0.04 0.05
104 62 109 118 96 1.34 1.23 1.36
(6) 12 +22 +6 +5 0.07 0.07 0.07
105 69 139 140 113 1.25 1.22 1.36
(3) 9 +24 + 11 +6 0.11 0.07 0.09

The mean critical crack size for the control group (0 cycles) is 53 11 [tm.
() Number of specimens in each group






















(b) 103cycles (150X)


(c) 104cycles (230X)


Figure 4.11.


(d) 105cycles (200X)


SEM images of fracture surfaces of group AR 3 specimens subjected to
cyclic loading from 103 to 105 cycles: (a) and (b) reveal semi-elliptical
cracks and (c) and (d) show triangular cracks.


low


(a) 0 cycles (150X)


S 200m I






49



(a) Air


0 10000 100000
Loading cycles


(b) Deionized water


0 1000 10000 100000
Loading cycles


Figure 4.12.


Crack size as a function of loading cycles for AR 3 specimens. The
legends in the graphs refer to the parameters used in Eqs. 3.2, 4.3, 4.5, 4.7
and in Table 4.6 and 4.7.























(a) 104cycles (100X)


(b) 105cycles (75X)

Figure 4.13. SEM images of fracture surfaces of group AR 3 specimens subjected to
cyclic loading for 104 and 105 cycles in deionized water.









For the AR 8 group, the critical crack sizes are summarized in Table 4.8. Crack

sizes as a function of loading cycles for group AR 8 specimens both tested in air and in

deionized water are shown in Fig. 4.14. Failure of all specimens initiated from surface

damage caused by either polishing or cyclic loading. The mean crack sizes of specimens

subjected to cyclic loading in ambient room temperature show no increase after 104 to 105

cycles, and no contact surface damage, i.e., ring or cone crack was observed in most of

the specimens. On the other hand, the strength of these bars tested in deionized water

decreased rapidly after 103 cycles, corresponding to a considerable increase in critical

crack size. In specimens subjected to loading for 104 tol05 cycles in deionized water, a

surface impression and subsurface damage zone can be clearly seen. The fracture

surfaces of group AR 8 specimens subjected to cyclic loading for 103 to 105 cycles in air

and in deionized water are shown in Figs. 4.15 and 4.16.

Table 4.8. Critical crack sizes of fractured group AR 8 specimens.

AR 8 0 cycles 103 cycles 104 cycles 105 cycles

Air (pm) 56 + 12 56 + 7 69 +13 71 + 8

Deionized water ([tm) 92 + 7 76 + 16 108 + 26 265 + 81

















1250

200

150

100

50

0


1000


10000


100000


Loading cycles


Figure 4.14. Crack size as a function of loading cycles for group AR 8 specimens tested
in air and in deionized water.


............. ............. A ir
*--- Deionized water












....... 4 .























(a) 104 cycles (65X)


(b) 105 cycles (100X)

Figure 4.15. SEM images of fracture surfaces of group AR8 specimens subjected to
loading for 104 and 105 cycles.



























(a) 104 cycles (200X)


(b) 105 cycles 120X

Figure 4.16. SEM images of fracture surfaces of group AR 8 specimens subjected to
loading for 104 and 105 cycles in deionized water.


-MMNW-J -""Opp-









Fatigue Failure Mechanisms

Representative SEM images of fracture surfaces for baria-silicate glass-ceramics

subjected to loading for 105 cycles in deionized water for each aspect ratio group are

shown in Figs. 4.17 (AR 3) and 4.21 (AR 8). Figure 4.17 shows the fracture surfaces and

critical crack size for an AR 3 specimen (aspect ratio 3/1). The fracture surface appears

to be tortuous and boundaries of crystallites can be clearly seen. Contact damage can be

seen at the top surface. A crushing zone occurred near the surface, including microcracks

propagating downward from the contact surface. A ring crack developed outside the

contact circle and propagated downward to form a cone crack. However, the failure of

this specimen did not start from the tip of a cone crack, as occurred in the AR 3

specimens tested in air, where the failure initiation sites were usually associated with the

cone crack without evidence of severe contact damage (Fig. 4.19a). For AR 3 specimens

tested in air, slight roughening of the contact surface of specimens subjected to 105

loading cycles was the only sign indicating damage from cyclic loading in addition to a

ring and cone crack formation. For AR 3 specimens tested in deionized water, an area of

microcracks propagating downward from the contact surface can be clearly seen in Fig.

4.18b. This contact damage zone was a failure initiation site for catastrophic fracture as

shown in Fig. 4.13b. The aqueous environment caused a change in the fracture initiation

site from the tip of a cone crack to surface contact damage by accelerating the crack

initiation and propagation rate at the contact surface. The wear surface was first observed

in group AR 3 specimens subjected to 104 cycles in deionized water. The damage was

more severe for specimens loaded for 105 cycles as shown in Fig. 4.19. Two principal

factors are associated with the formation of microcracks within the surface and

subsurface damage zone, stress corrosion fatigue and residual stress. Stress corrosion









fatigue can easily occur at the surface damage area during cyclic loading because there is

much material exposed to the environment and this process enhances the propagation of

the microcracks. The formation of these microcracks also increases the compliance of

the material in the damage zone. As a result, the residual stress develops from a

mismatch between the elastic modulus of the damage zone and the surrounding area.

This residual stress also enhances the formation of microcracks within the contact

damage area. The strain mismatch between the glass and crystal phases associated with

cyclic loading can also cause residual tensile stress during the unloading half-cycle (18).

Hill (58) observed the fracture path of group AR 3 specimens propagating through

the glass matrix. He suggested that crack deflection is a toughening mechanism for this

material. Failure of specimens used in a previous study (58) occurred during a quasi-

static four-point loading test. Therefore, failure origins of those specimens were caused

by flaws produced during grinding and polishing. In the present study the critical crack

propagated from a cone crack that developed during blunt indentation loading, or that

originated from the damage zone beneath the contact surface. The propagation path for

the cone crack appeared to progress straight downward and likely through the crystal

phase (Fig.4.18a), following the path perpendicular to the direction of the maximum

tensile stress. Therefore, a transgranular fracture mode was observed after cyclic loading

in group AR 3 specimens. Analytical results from previous studies suggest that failure

from a cone crack should initiate from one of the two opposite points on the base rim of

the cone (53-54). However, we found that only a few specimens failed from the tip of the

cone crack. Instead, most of the AR 3 specimens failed from a triangular crack

developed during fast fracture.

























0~


0,






0 0


C~j~









C- ~-

C~









Figure 4.21 illustrates the fracture surfaces of the AR 8 group (aspect ratio 8/1).

The surfaces appear to be more tortuous, but with sharp deflection angles instead of the

smoother curves seen in the fracture path for the AR 3 group (Fig. 4.22a). Hill (58)

suggested that the crack path of this group was transgranular through the baria-silicate

crystals indicating that the fracture propagated along a new path of least resistance rather

than around the crystals as seen in the AR 3 group. He also suggested crack bridging as

an additional toughening mechanism in addition to crack deflection for this group.















(a) (b)

Figure 4.18. SEM image of fracture surfaces of group AR3 specimens subjected to
loading for 105 cycles in deionized water (2000X): (a) cone crack-related
failure (higher magnification of Fig. 4.13a); (b) surface damage-related
failure (higher magnification of Fig. 4.17).

















(a)

















(b)


Figure 4.19.


SEM images of surface contact damage in AR 3 specimens: (a) after 105
cycles in air; (b) after 105 cycles in deionized water.









In the present study, transgranular fracture occurred through the crystal phase in the

crushing zone beneath the contact surface and other areas (Fig.4.21), suggesting the same

fracture path and the toughening mechanism as that which occurred in the quasi-static

loading failure. We did not observe failure originating from a cone crack in AR 8

specimens. Based on the fracture surface analysis of group AR 8 specimens tested in air,

the mean crack size increased slightly in proportion to the increase in numbers of loading

cycles. However, no subsurface damage was detected in those specimens except for the

slight rough surface observed after 105 cycles (Fig.4.20a). In contrast, group AR 8

specimens tested in deionized water showed a significant effect of stress corrosion fatigue

(Fig. 4.20b). The mean strength values of these bars tested in deionized water decreased

rapidly after 103 cycles, corresponding to a significant increase in critical crack sizes. In

specimens subjected to loading for 104 tol05 cycles (Fig. 4.21) in deionized water, a

surface impression and subsurface damage zone can be clearly seen. Areas of

microcracks propagating downward from the contact surface were observed and are

shown in Fig. 4.22. The dominant fracture path is transgranular (Fig. 4.22a), which is

similar to that observed after quasi-static loading failure (58).

The mechanisms of microcrack formation in group AR 8 specimens were similar to

those described earlier for group AR 3 specimens. However, the effect of an aqueous

environment was significant for group AR 8 specimens. Because of its high fracture

toughness (2.11 + 0.16 MPa.m1/2), group AR 8 specimens are more resistant to crack

propagation compared with the glass and AR 3 groups, which we can verify for the

specimens tested in air. However, the aqueous environment accelerated the crack

propagation rate at the surface contact area in both AR 3 and AR 8 groups. Thus, we









observed severe surface damage in specimens tested in deionized water. This resulted in

a significant decrease in the mean strength for group AR 8 specimens because the failure

initiation site for group AR 8 specimens was at the contact surface. For group AR 3

specimens, even though we observed severe surface damage, the failure initiation sites of

specimens tested in air were different from those observed in specimens tested in

deionized water. Because of the change of failure initiation sites from a cone-crack

related flaw to contact surface damage, we did not observe a significant decrease in

strength for group AR 3 specimens tested in air and in deionized water after 105 cycles.

In this study, the contact fatigue damage patterns of glass-ceramic specimens are

similar to the wear process of silicon nitride observed in a previous study (43). The wear

mechanisms of silicon nitride with a fine surface finish can be divided into three stages.

In stage I, roughening of the contact surface occurs because of the damage to the grain

boundaries. In stage II, ring and cone cracks form and propagate. In stage III, a

significant amount of material is lost from the region outside the contact area. If the

surface finish is poor, stage II can occur before stage I (43).

Stage I occurs because of the viscoelastic behavior of the glass phase that is

distributed along the grain boundaries of silicon nitride during the sintering process. The

delayed response of the glass phase allows tensile stress to develop during unloading

because of the strain mismatch between the two phases. Eventually, the cyclic tensile

stress in the glass phase causes grain boundary fracture. The delayed response of the

glass phase as well as the large loads also causes shear stress to develop just outside the

contact region. This distribution of shear stress can enhance crack initiation near the

contact circle. In this study, roughening of the contact surface was observed on









specimens subjected to cyclic loading in group AR 3 and group AR 8 specimens tested in

deionized water (Fig.4.23). The glass matrix in baria-silicate glass-ceramics behaves as a

site for the residual tensile stress to occur. The protruding crystals and sites associated

with lost crystals are clearly seen in Fig. 4.23.

Stage II occurs because of the high tensile stress at the edge of the contact region in

the Hertzian stress field and the existence of the surface cracks. In this study, a ring and

cone crack were observed in most of the AR 3 specimens subjected to a sufficient

number of loading cycles (Fig. 4.19). For group AR 8 specimens, no ring or cone crack

was observed because of the dissipation of the contact stress that can decrease the

intensity of the Hertzian stress fields (Fig. 4.20).

Stage III involves the removal of material from the surface and may resemble the

spallation caused by subsurface cracks that propagate to the surface (59). For stage III

wear, cracks start at the contact surface and propagate downward a certain distance and

then progress upwards toward the surface (43, 59). The reason for the change in the

direction of the propagating cracks may be the change in stress fields when there is the

presence of cracks in the material structure. High tensile stress fields are observed close

to the crack tip (59). A high tensile stress field at the surface and at the crack tip can

cause the cracks to propagate toward the surface. The upward-curving cracks eventually

result in a loss of material from the region outside the contact region. In the present

study, no stage III wear was observed in the groups AR 3 and AR 8 tested in air. In

contrast, severe surface damage caused from material loss was observed in group AR 3

and AR 8 specimens tested in deionized water (Fig.4.24). Thus, stress corrosion has a

significant effect at this stage, by increasing the rate of the crack initiation and









propagation in baria-silicate glass-ceramics. This process results in the creation of

strength-limiting flaws that cause significant decreases in the strength of glass-ceramic

materials.

In summary, cyclic loading creates strength-limiting flaws that can affect the

fracture susceptibility of ceramic materials. For materials with low toughness (glass and

AR 3), a ring and cone crack were the dominant flaws that caused the decrease in the

mean strength of these materials. For the material with a higher toughness (AR 8),

surface flaws created during contact fatigue were strength limiting. Stress corrosion

enhanced the crack propagation rate and caused a significant decrease in strength of the

AR 8 group. The contact fatigue damage patterns observed in this study were similar to

the wear process reported in a previous study (43), including three stages of progressive

surface and subsurface damage.

In this study, the glass-ceramics were produced with two aspect ratios using

controlled heat treatment regimens to study the effect of microstructure on fatigue failure

mechanisms. We selected a baria-silicate composition because there should be little

residual stress caused by elastic mismatch or a thermal expansion mismatch during the

cooling process. Therefore, we minimized the effect of the residual stress caused from

these mismatches and we observed the actual cyclic behavior of baria-silicate glass-

ceramics.

Previous studies showed that the microstructure of a micaceous glass-ceramic

system was a major factor in controlling the damage modes under Hertzian contact

loading (9, 10). By heat-treating glass specimens, the investigations yielded a series of

glass-ceramics with microstructures that contained fine (0.3 |tm x 1 [tm) and coarse









platelets (1.2 |tm x 8 [tm). Contact damage patterns from blunt indentation of bonded-

interface specimens can be classified into two damage modes: (1) ring and cone cracks

that are dominant in fine microstructures, and (2) subsurface deformation or quasi-

plasticity that occurred in coarse microstructures. The authors concluded that

microstructural coarsening of micaceous glass-ceramics produced a brittle-plastic

transition.






n ,'i


Figure 4.20. Images of surface contact damage in group AR 8 specimens (a) after 105
cycles in air; (b) after 105 cycles in deionized water.


Contact damage






















'hi


0l
0~


0 0
o




























S0 ,
zs
n o









a M








I-I
O41
O^
x



































(a) 104 cycles


(b) 105 cycles


Figure 4.22. SEM images of fracture surfaces of group AR8 specimens subjected to
loading for 104 and 105 cycles in deionized water (3000X).


s:.



















(a)

















(b)









Figure 4.23. SEM images of the cyclically loaded surfaces of baria-silicate glass-
ceramics (a) AR 3 and; (b) AR 8 specimens tested in deionized water.




















(a)














(b)








Figure 4.24. SEM images of the cyclically loaded surfaces of baria-silicate glass-
ceramics (a) AR 3 and; (b) AR 8 specimens tested in deionized water.









In the present study, we observed the same fracture patterns as reported earlier, but

some additional explanation is necessary. Relative to the critical crack configurations,

we observed that the fracture paths progressed through the contact damage in specimens

with a high aspect ratio (AR 8), or near the ring crack in glass specimens in a similar

manner as was reported in other studies (9, 10). Previous investigators (9, 10) did not

identify whether the failure origins occurred from the cone cracks or from contact

damage. Based on fractographic analysis of our specimens, we found three different

crack configurations that cannot be clearly described by a simple cone-crack model (53).

The second point is that the authors (12, 13, 60) also suggested that the Hertzian test

simulates oral loading conditions by using contact loads that are similar to intra-oral

occlusal loads, and an indenter radius that is comparable to the cuspal radii of teeth.

Indentation loads of up to 3000 N were used, which were very high compared to normal

occlusal loads (89-890 N) (61). Furthermore, for indented specimens, the tungsten

carbide indenters that they used have a much higher elastic modulus than that of tooth

enamel. As a result, the authors suggested that a quasi-plastic zone beneath the contact

area was a failure initiation site for alumina and silicon carbide materials (27). This

quasi-plastic zone developed in a compressive stress field caused from Hertzian contact

loading. In the present study, the cyclic contact load of 200 N was low compared with

that used in the previous studies. The damage zone was confined near the surface. It is

evident that microcracks formed in that damage zone originated and propagated

downward from the contact surface, not from the compressive zone beneath the contact

area.









Clinical failure mechanisms are currently a topic of a major concern. Although

some investigators prefer a blunt indentation test since it can characterize materials

properties under clinically relevant conditions, others argue that indentation damage is

typically not associated with clinical failures of dental prostheses (9-12, 62, 63). From a

fractographic perspective, failure initiation sites of failed ceramic prostheses appear to

occur at the cementation surface or the interface between core and veneer (14, 15), not

within occlusal surface. In spite of this difference, there is little doubt that contact

damage can occur during mastication, because wear of the occlusal surfaces can be

generally observed in ceramic prostheses. Even though this damage may not lead to

immediate failure, a decrease in the critical stress for failure can be expected. A partial

loss of ceramic prosthesis structure (chipping) caused by contact damage is also possible

as a treatment failure because of compromised esthetics and function. To obtain useful in

vitro information that is relevant to clinical failure conditions, well-designed experiments

under variable loading parameters and simulated clinical conditions are recommended for

future studies of the failure behavior of dental ceramics.














CHAPTER 5
CONCLUSIONS

The conclusions of this study are as follows:

(1) Cyclic loading decreases the strength of baria-silicate glass and glass-ceramics.

(2) Four different crack geometries were identified and associated with failure

mechanisms. For glass specimens, the decrease in fracture stress after 103 cycles was

associated with cone crack formation. No significant additional reduction was found

after loading for 103 cycles and 105 cycles of specimens tested both in air and in

deionized water.

(3) For the glass-ceramic groups with a crystal aspect ratio of 3/1 tested in air, the

decrease in fracture stress occurred initially after 104 cycles and was associated with the

formation of cone cracks. The decrease in fracture stress observed after 103 cycles for

specimens tested in deionized water, suggests an acceleration of crack propagation

because of stress corrosion.

(4) For the glass-ceramic groups with a crystal aspect ratio of 8/1, we did not

observe failure originating from a cone crack in any specimens. We observed a gradual

decrease in fracture stress in specimens subjected to loading in air for 103 to 105 cycles.

A statistically significant decrease was found in specimens cyclically tested in deionized

water as compared to those tested in air. A reduction of approximately 50% in fracture

stress was found for specimens subjected to loading for 105 cycles in deionized water.

(5) Comparisons between methods used to determine the possible crack sizes were

made. The geometric constants obtained from the surface area analyses (Equations 3.2









and 4.7) are more rational than those obtained from the cone-crack analysis models

(Equations 4.3 and 4.5). However, an error is associated with the methods that are used

to determine the surface area of critical cracks in this study because the curvature of a

cone crack was disregarded to simplify the calculation.

(6) Control of microstructure by heat treatment can greatly affect the strength and

failure mechanisms of glass-ceramics during cyclic fatigue testing in air. The failure

initiation mechanisms change from one occurring at the base of the cone-crack in baria-

silicate glass and glass-ceramics with a crystal aspect ratio of 3/1 (AP 3) to that resulting

from surface contact damage in glass-ceramics with an crystal aspect ratio of 8/1 (AR 8).

(7) An intergranular fracture path was observed in glass-ceramics with an aspect

ratio of 3/1, suggesting that the crack deflection is a major toughening mechanism. For

an aspect ratio of 8/1, a transgranular fracture mode was dominant and crack bridging and

crack deflection contributed to toughening.

(8) Stress corrosion fatigue accelerated the crack propagation rate in baria-silicate

glass-ceramic specimens, especially in specimens with an aspect ratio of 8/1. The mean

crack sizes were significantly larger in AR 8 specimens subjected to cyclic loading in

deionized water, compared with those subjected to the same numbers of cycles in air.















REFERENCES


1. Walton JN, Gardner FM, Agar JR (1986). A survey of crown and fixed partial
denture failures: Length of service and reasons for replacement. JP1i thet Dent
56:416-21.

2. Foster LV (1990). Failed conventional bridge work from general dental practice:
Clinical aspects and treatment needs 142 cases. Br Dent J 168:199-201.

3. Moffa JP, Lugassy AA, Ellison JA (1988). Clinical evaluation of a castable
ceramic material (abstract). JDent Res 67:118.

4. White SN (1993). Mechanical fatigue of a feldspathic dental porcelain. Dent
Mater 9:260-4.

5. White SN, Zhao XU, Yu Z, Li ZC (1995). Cyclic mechanical fatigue of a
feldspathic dental porcelain. Int JProsthodent 8:1-8.

6. White SN, Li ZC, Yu Z, Kipnis V (1997). Relationship between static chemical
and cyclic mechanical fatigue in a feldspathic porcelain. Dent Mater 13:103-10.

7. Sobrinho LC, Cattell MJ, Glover RH, Knowles JC (1998). Investigation of the dry
and wet fatigue properties of three all-ceramic crown systems. Int JP1i,,,th,,1,unt
11:255-62.

8. Morena R, Beaudreau GM, Lockwood PE, Evans AL, Fairhurst CW (1986).
Fatigue of dental ceramics in a simulated oral environment. JDent Res 65:993-7.

9. Peterson IM, Wuttiphan S, Lawn BR, Chyung K (1998). Role of microstructure on
contact damage and strength degradation of micaceous glass-ceramics. Dent Mater
14:80-9.

10. Peterson IM, Parjares A, Lawn BR, Thompson VP, Rekow ED (1998). Mechanical
characterization of dental ceramics by Hertzian contacts. JDent Res 77:589-602.

11. Cai H, Kalceff MS, Lawn BR (1994). Deformation and fracture of mica-containing
glass-ceramics in Hertzian contacts. J Mater Res 9:762-70.

12. Cai H, Kalceff MS, Hooks BM, Lawn BR (1994). Cyclic fatigue of a mica-
containing glass-ceramics at Hertzian contacts. J Mater Res 9:2654-61.









13. Lawn BR, Deng Y, Thompson VP (2001). Use of contact testing in the
characterization and design of all-ceramic crownlike layer structures: a review. J
Prosthet Dent 86:495-510.

14. Kelly JR, Giordano R, Pober R, Cima MJ (1990). Fracture surface analysis of
dental ceramics: clinically failed restorations. Int JPi r,iihe,,,unt 3:430-40.

15. Thompson JY, Anusavice KJ, Naman A, Morris HF (1994). Fracture surface
characterization of clinically failed all-ceramic crowns. JDent Res 73:1824-32.

16. MacDowell JF (1965). Nucleation and crystallization of barium silicate glasses.
Proc Brit Ceram Soc 3:229-40.

17. Freiman SW, Onoda GY, Pincus AG (1974). Mechanical properties of 3BaO-
5Si0O2 glass. JAm Ceram Soc 57:8-12.

18. Suresh S (1998). Fatigue of materials. Cambridge University Press, Cambridge,
U.K.

19. Ritchie RO (1988). Mechanisms of fatigue crack propagation in metals, ceramics
and composites: role of crack tip shielding. Mater Sci Eng Al 03:15-28

20. Ritchie RO, Dauskardt RH, Yu W (1990). Cyclic fatigue-crack propagation,
stress-corrosion, and fracture-toughness behavior in pyrolytic carbon-coated
graphite for prosthetic heart valve applications. JBiomedMater Res 24:189-206.

21. Ritchie RO, Gilbert CJ, McNaney JM (2000). Mechanics and mechanisms of
fatigue damage and crack growth in advanced materials. Int JSolids Struc 37:311-
29.

22. Lathabai S, Rodel J, Lawn BR (1991). Cyclic fatigue from frictional degradation at
bridging grains in alumina. JAm Ceram Soc 74:1340-8.

23. Dauskardt RH (1993). A frictional wear mechanism for fatigue-crack growth in
grain bridging ceramics. Acta Metal Mater 41:2765-81.

24. Lawn BR, Padture NP, Guiberteau F, Cai H (1994). A model for microcrack
initiation and propagation beneath Hertzian contacts in polycrystalline ceramics.
Acta Metal Mater 42:1683-93.

25. Argon AS (1982). Fracture in compression of brittle solids. National Materials
Advisory Board Report NMAB-404, National Academy Press, Washington, DC.

26. Ewart R, Suresh S (1987). Crack propagation in ceramics under cyclic loads. J
Mater Sci 22:1173-92.

27. Lawn BR (1998). Indentation of ceramics with spheres: a century after Hertz. J
Am Ceram Soc 81:1977-94.






75


28. Roesler FC (1956). Brittle fractures near equilibrium. Proc Phys Soc London
B69:981-92.

29. Lawn B (1993). Fracture of brittle solids. Cambridge University Press, Cambridge,
U.K.

30. Lee SK, Lawn BR (1998). Role of microstructure in Hertzian contact damage in
silicon nitride: strength degradation. JAm Ceram Soc 81:997-1003.

31. Wiederhorn SM, Bolz LH (1970). Stress corrosion and static fatigue of glass. J
Am Ceram Soc 53:543-8.

32. Michalske TA, Freiman SW (1982). A molecular interpretation of stress corrosion
in silica. Nature 295:511-2.

33. Ebrahimi ME, Chevalier J, Fantozzi G (1999). Slow crack-growth behavior of
alumina ceramics. JMater Res 15:142-7.

34. Chevalier J, Olagnon C, Fantozzi G (1999). Subcritical crack propagation in 3Y-
TZP ceramics: static and cyclic fatigue. JAm Ceram Soc 82:3129-38.

35. Ebrahimi ME, Chevalier J, Fantozzi G (2000). Environmental effects on crack
propagation of alumina ceramics. Ceram Eng Sci Proc 21:385-96.

36. Asoo B, McNaney JM, Mitamura Y, Ritchie RO (2000). Cyclic fatigue-crack
propagation in sapphire in air and simulated physiological environments. JBiomed
Mater Res 52:488-91.

37. Tanaka H, Bando Y (1990). Atomic crack tips in silicon carbide and silicon
crystals. JAm Ceram Soc 73:761-3.

38. Cook RH (1986). Crack propagation thresholds: a measure of surface energy. J
Mater Res 1:852-60.

39. Thomson R (1990). The molecular wedge in a brittle crack: a simulation of
mica/water. J Mater Res 5:524-34.

40. Hertzberg RW (1995). Deformation and fracture mechanics of engineering
materials. John Wiley & Sons, Inc., New York, U.S.A.

41. Choi G, Horibe S (1995). The environmental effect on cyclic fatigue behavior in
ceramic materials. JMater Sci 30:1565-9.

42. Jacobs DS, Chen IW (1995). Cyclic fatigue in ceramics: a balance between crack
shielding accumulation and degradation. JAm Ceram Soc 78:513-20.

43. Chen Z, Cuneo JC, Mecholsky JJ, Hu S (1996). Damage processes in Si3N4
bearing material under contact loading. Wear 198:197-207.









44. Paris PC, Gomez MP, Anderson WP (1961). A rational analytic trend of fatigue.
Trend in Engineering 13:9-14.

45. Sylva LA, Suresh S (1989). Crack growth in transforming ceramics under cyclic
tensile loads. JMater Sci 24:1729-38.

46. Dauskardt RH, Marshall DB, Ritchie RO (1990). Cyclic fatigue-crack propagation
in magnesia-partially-stabilized zirconia ceramics. JAm Ceram Soc 73:893-903.

47. Jones DW, Sutow EJ (1987). Stress corrosion failure of dental porcelain. Br
Ceram Trans J 86:40-3.

48. Adam R, McMillan PW (1977). Static fatigue in glasses. JMat Sci 12:643-57.

49. Sherrill CA, O'Brien WJ (1974). Transverse strength of aluminous and feldspathic
porcelain. JDentRes 53:683-90.

50. Southan DE, Jorgensen KD (1974). The endurance limit of dental porcelain. Aust
Dent J 19:7-11.

51. Irwin GR (1962). Crack-extension force for a part-through crack in a plate. J
Appli Mech 29:651.

52. Mecholsky JJ, Passoja DE, Feinberg-Ringel KS (1989). Quantitative analysis of
brittle fracture surfaces using fractal geometry. JAm Ceram Soc 72:460-5.

53. Lawn BR, Wiederhorn SM, Johnson HH (1975). Strength degradation of brittle
surfaces: blunt indenters. JAm Ceram Soc 58:428-32.

54. Evans AG (1973). Strength degradation by projectile impacts. JAm Ceram Soc
56:405-9.

55. Murakami Y (1985). Analysis of stress intensity factors of mode I, II and III for
inclined surface cracks of arbitrary shape. Eng Frac Mech 22:101-14.

56. Widjaja S, Ritter JE, Jakus K (1996). Influence of R-curve behavior on strength
degradation due to Hertzian indentation. JMater Sci 31: 2379-84.

57. Brown WF, Strawley JE (1967). Plane strain crack toughness testing of high
strength metallic materials. ASTM Special technical publication No. 410: 89-129.

58. Hill TJ, Mecholsky JJ, Anusavice KJ (1995). Fracture toughness versus
crystallization in baria-silicate glass-ceramics. Am Assoc Dent Res Abstract 1657.

59. Chao L, Shetty DK, Adair JH, Mecholsky JJ (1996). Development of silicon
nitride for rolling-contact bearing applications a review. JMater Edu 17:245.

60. Jung YG, Peterson IM, Kim DK, Lawn BR (2000). Lifetime-limiting strength
degradation from contact fatigue in dental ceramics. JDent Res 79:722-31.






77


61. Anusavice KJ (1996). Philllip's science of dental materials, 10th edition. W.B.
Saunders Company, Philadelphia, USA.

62. Kelly JR, Tesk JA, Sorensen JA (1995). Failure of all-ceramic fixed partial
dentures in vitro and in vivo: analysis and modeling. JDent Res 74:1253-8.

63. Kelly JR (1999). Clinically relevant approach to failure testing of all-ceramic
restorations. JP /,ithet Dent 81:652-61.















BIOGRAPHICAL SKETCH

Kallaya Suputtamongkol was born in Bangkok, Thailand, on December 15, 1968.

After graduation from a local high school in 1984, she entered Mahidol University and

earned a Doctor of Dental Surgery degree in 1990. She continued her graduate study at

Mahidol University and obtained a Master of Science degree in prosthodontics in 1997.

After graduation, she has started working at the Faculty of Dentistry, Mahidol University,

as an instructor in the Prosthodontics Department. She has continued her education by

entering the Ph.D program at the University of Florida, College of Engineering, in 1998.