- Email: [email protected]
Interfacial fracture toughness between bovine cortical bone and cements
Biomaterials 24 (2003) 1159–1166
Interfacial fracture toughness between bovine cortical bone and cements P. Lucksanasomboola,c, W.A.J Higgsb,c, R.J.E.D Higgsa,b, M.V Swainb,c,* a Faculty of Medicine, University of Sydney, Australia Department of Mechanical & Mechatronic Engineering, University of Sydney, Australia c Biomaterials Science Research Unit, Faculty of Dentistry, University of Sydney, Suite G11, National Innovation Centre, Australian Technology Park, Eveleigh, NSW 1430, Australia b
Received 10 July 2001; accepted 20 July 2002
Abstract To evaluate the bonding strength of the interfaces within the cemented arthroplasty system, various mechanical tests have been used. Conventional push-out and pull-out tests cannot reveal the actual bonding property of the interface because of the significant influence of surface roughness on the measured adhesion and the failure to account for the mismatch of elastic modulus across the interface. An alternative fracture mechanics approach, which considers the mix of opening and shear modes of the crack tip loading associated with the testing system and the elastic mismatch of materials across the interface, was used to evaluate the bonding ability of various cements. The four-point bend interfacial delamination test by Charalambides et al. (J. Appl. Mech. 56 (1989) 77; Mech. Mater. 8 (1990) 269) was used to quantify the bonding ability of cements. This method is arguably more suitable since the applied loading mode is comparable to the nature of loading within the prosthetic system, which is primarily bending. The bovine bone specimens were polished to mirror finish to eliminate bonding by mechanical interlocking. The results revealed minimal bonding for the conventional bone cement (PMMA) whereas substantial bonding was evident for the glass-ionomer cements tested. However, only the conventional glass-ionomer cements showed evidence of bonding on testing, while the resin-modified glass-ionomer cement (poly-HEMA) did not. The latter appeared to debond before testing because of excessive expansion stresses associated with swelling in water. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Interfacial fracture toughness; Critical strain energy release rate; Phase angle; Bone cements; Glass ionomer cements
1. Introduction Debonding of the cement-metal interface has been implicated as a major site of failure initiation [1,2] in total joint replacement. There have been many attempts to improve the efficacy of bonding by such means as improved cementing techniques, and the addition of bonding agents to existing cement systems. In order to evaluate the improvement in bonding, conventional push-out and pull-out tests [3–7] and fracture mechanics approaches [8–10] have been used. Prior investigations studied the bonding between bone and bone cement using conventional strength methods such as push-out and pull-out tests [3–7]. The resultant *Corresponding author. Tel.: +61-2-9351-1814; fax: +61-2-93511815. E-mail address: [email protected] (M.V. Swain).
failure characteristics were unable to reveal an actual bonding property of the interface and inter-comparison has proved exceptionally difficult because of uncertain types of failure (bone fracture, cement fracture and/or mixed) (Table 1). Various types of failures occurred in push-out and pull-out tests because of the mix of loading modes which depend on the loading geometry and the mismatch of the elastic modulus of the two materials. Wang et al. [11] conducted bimaterial bonding strength tests using different techniques and found that push-out and pull-out tests were sensitive to specimen configuration and had less reproducibility due to the material mismatch across the interface. This mismatch resulted in the complex stress distribution along the interface. In subsequent studies, a fracture mechanics approach was introduced with the aim of better evaluating debonding at the interface. Tests of this type are more colloquially referred to as interfacial
0142-9612/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 1 4 2 - 9 6 1 2 ( 0 2 ) 0 0 4 6 4 - 7
P. Lucksanasombool et al. / Biomaterials 24 (2003) 1159–1166
1160 Table 1 Previous interfacial strength studies Studies
Test methods
Failure sites
Interfacial strength
Krause et al. (1982)
Tensile pull-out test
2.7271.42 MPa
Krause et al. (1982)
Shear-cylindrical specimens test
Through cancellous-cement interface (mo cleaning; fingerpacked) Through cement (high-intensity lavage; pressure injected) Not be described (no cleaning; finger-packed)
Kusleika and Stupp (1983)
Tensile pull-out test
Askew et al. (1984)
Tensile test
Macdonald et al. (1993) Mann et al. (1997)
Shear pull-out test Tensile pull-out test
Not be described (high-intensity lavage; pressure injected) Cement fracture (most common) Mixed fracture Bone fracture (less common) No penetration Penetration cement was pulled out from bone Bone fracture Cement fracture Not be described Through cancellous-cement interface (40 specimens) Through composite (27 specimens) Through bone adjacent to the interface (4 specimens)
fracture toughness tests [12–19]. This conditioned approach is better able to reveal the interfacial bonding as it considers both the loading mode and the material mismatch within the testing procedures [11]. Using a fracture mechanics approach, the mixture of materials across the interface and the testing procedure result in a mixed opening and shear loading mode developing at the crack tip that is characterized by the phase angle, c; which is equal to the ratio of the shearing (mode II) to opening (mode I) stress intensity factors. Therefore, in pure mode I, the phase angle is 01 and is 901 in pure mode II. With differing specimen geometry and applied loading, the resultant interfacial toughness varies because the phase angle will change (Fig. 1). The resultant interfacial fracture toughness can be described in terms of the stress intensity factor, K; or the critical strain energy release rate, G: However, the bimaterial stress intensity factor and phase angle c are scale sensitive [20]. Therefore, it is more convenient for comparison to consider the interface fracture resistance locus in terms of trends in critical strain energy release rate G; which is scale insensitive [15]. The four-point bend delamination test (Fig. 1), which produces stable crack propagation, was developed for measuring the fracture resistance of bimaterial interfaces by Charalambides et al. [15,16]. The test is appropriate for testing the interfaces between cortical bone and bone cement because the loading mode is relevant to one of the loading modes (bending) within the prosthetic system. In this study, the four-point bend delamination test was used to measure the bonding ability between bovine cortical bone and three different types of cements, consisting of the conventional bone cement (PMMA), two conventional glass ionomer cements
8.5074.28 MPa 9.5574.06 MPa 41.9975.89 MPa 6–9 MPa 5–8 MPa 4–6 MPa Cannot be determined
3.2–7.3 MPa 0.9870.69 MPa 1.6870.66 MPa 1.4970.7 MPa
(FUJI II & IX), and the novel resin-modified glass ionomer cement (S430). 2. Materials and methods 2.1. Four-point bend delamination test: theoretical framework The specimen consists of a notched, bimaterial flexural beam (Fig. 1). A central notch exits through the thickness of one material and a symmetrical precrack is induced along the interface. When the specimen is loaded in four-point bending, a steady-state scheme can be obtained within the inner loading lines [15] and at least when the interface crack length significantly exceeds the thickness of the upper layer of the beam. The steady-state value, Gss ; can be determined analytically by simply comparing the difference in the strain energy in the uncracked and cracked beam. From Euler–Bernoulli beam theory and plain strain conditions, the strain energy per unit cross-section, U; can be expressed in terms of the applied moment M as ð1 n2 ÞM 2 ; 2EI where I is the second moment of area per unit width, E is the elastic modulus of the beam, and n is the Possion’s ratio. When the equation is applied to the bimaterial for the steady-state value, Gss ; which is the difference in the strain energy in the uncracked and cracked beam, the equation is M 2 ð1 n22 Þ 1 l Gss ¼ ; 2E2 I2 Ic U¼
P. Lucksanasombool et al. / Biomaterials 24 (2003) 1159–1166
1161
Fig. 1. Schematic illustration of various specimen configurations with their phase angles.
where the subscript 2 describes the properties of the unnotched layer, the subscript c refers the property of the composite beam and l is l¼
E2 ð1 n21 Þ : E1 ð1 n22 Þ
where Z is equal ( 3 1 Z¼ 2 ðh2 =hÞ3 "
l 3 3 ðh1 =hÞ þ lðh2 =hÞ þ 3lðh1 h2 =h2 Þðh1 =h þ lh2 =hÞ1
The subscript 1 describes the properties of the notched layer of composite beam. Ic can be calculated as the equation: Ic ¼ h31 =12 þ lh32 12 þ lh1 h2 ðh1 þ h2 Þ2 =4ðh1 þ lh2 Þ; where hn is the height or thickness of each layer and I2 can be expressed as
#) :
The steady-state strain energy release rate, Gss ; with the unit of J/m2 has also been analysed in terms of the effects of the thickness and moduli of each material constructing the bimaterial specimen, the frictional effects, the residual stress effects, and also the interpretation of the load–displacement curves [15,16,19].
I2 ¼ h32 =12:
2.2. Specimen preparation and testing
Noting that the moment per unit width M ¼ Pl=2b; with P being the total load, b is the width of the beam, and l is the space between inner and outer loading lines, the strain energy release rate turns into:
Whole bovine tibias of unknown age and post mortem period were used. The bovine bones were soaked with normal saline and wrapped with normal saline soaked gauze. They were kept in a plastic bag and stored at 201C before processing. In order to sustain less variability of the bone specimens [21,22], cortical bone mainly from the posterior aspect of bovine tibias was cut and milled to rectangular bars of size 2.5 10.0 25.0 mm3. The longitudinal axis of the bars was parallel to the longitudinal axis of the bone. The intramedullary aspect of the specimens, which was bonded with the cement, was sequential polished with finer diamond powders until a mirror finish was attained. The bone specimens were maintained wet in saline during all stages of tissue preparation.
E2 Gss h3 b2 =P2 l 2 ð1 n22 Þ ¼ ð3=2Þf1=ðh2 =hÞ3 l=fðh1 =hÞ3 þ lðh2 =hÞ3 þ 3lðh1 h2 =h2 Þ 1 h1 =h þ lh2 =h gg: When the constant value in the equation is replaced with Z, the equation for the steady-state strain energy release rate, Gss ; can be rewritten as Gss ¼
ZP2 l 2 ð1 n22 Þ ; E 2 b2 h3
1162
P. Lucksanasombool et al. / Biomaterials 24 (2003) 1159–1166
Cements used in this study were a conventional bone cement, PMMA (CMW), a conventional glass ionomer cement (FUJI II), a high viscosity conventional glass ionomer cement (FUJI IX), and a resin-modified glass ionomer cement (S430). Cements were mixed manually with the powder/liquid ratio and mixing instructions according to each manufacturer’s recommendations. To produce the bimaterial specimens, the cement was packed into the 5.0 10.0 25.0 mm3 mould over the sponge dried bone bar and pressurized until the cement set. Eleven bimaterial specimens for each cement group were made. The bimaterial specimens were cleaned and ground to remove the excessive cement then they were kept at 371C normal saline for 1 week. The bimaterial specimens were notched through the cement layer using a low-speed diamond blade saw and precracked in a narrow (o5 mm) three-point bending jig, together with simultaneous observation using an optical microscope, to limit the extension of the precracks before testing. The specimens were loaded in four-point bending with the loading rate equal to 0.1 mm/min using a Shimadzu testing machine. An optical microscope was used to observe crack front extension on both sides of the introduced notch. Observation of the fracture surfaces was performed using a scanning electron microscope (SEM). A thin carbon coating was applied to the fracture surfaces of the specimens to minimize electrostatic charging. Some bovine bone specimens as well as cement bars fabricated from the same mould were also tested for the
moduli in the same storage and testing conditions as the bimaterial.
3. Results The observed crack extension from the initial notch revealed two typical forms, symmetrical and asymmetrical crack extensions (Fig. 2). For the bimaterial specimens prepared from FUJI II and IX, the observed crack extension phenomena can be related well with the loading. In symmetrical crack extension, the load increased linearly until the crack initiated whereupon the load remained almost constant as indicated by the plateau portion of the load–displacement curve, which showed less inclination than the other portions. The load increased linearly again after the crack propagated beyond the inner loading points (Fig. 3). The crack extension in this case was steady as can be seen on the unload–reload curve (Fig. 4). In case of asymmetrical crack extension, the loading behaviour was the same, but having two shorter plateau portions corresponding with the crack extension on each side (Fig. 4). For the specimens prepared from CMW and S430, the crack extension occurred immediately after loading in some specimens, whereas, in the others, no indication of crack extension was evident from the force–displacement curves despite application of significantly high loads and displacements. In the specimens, which displayed no obvious crack extension during loading,
Fig. 2. Micrographs of crack extension in Charalambides’ test: (a) symmetrical crack extension, (b) asymmetrical crack extension.
P. Lucksanasombool et al. / Biomaterials 24 (2003) 1159–1166
300 CMW-Bone FUJI II-Bone FUJI IX-Bone S430-Bone
250
Load (N )
200 150 100 50 0 0
0.1
0.2 Displacement (mm.)
0.3
0.4
Fig. 3. The load–displacement diagram from Charalambides’ test of various cement-bone bimaterials. The plateau portion (Fuji II and Fuji IX bimaterials) reveals interfacial bonding.
1163
it was found that both layers were completely detached from each other. However, upon SEM observation, the bone surface of S430 specimens revealed cohesive failure (Fig. 5) of cement material similar to other GICs whereas there was no cement attached to the bone for CMW specimens. For the calculation of the strain energy release rate, the plateau loads, which were determined by the obvious changes of the inclination of the load–displacement, were averaged. The calculated strain energy release rate, Gss ; of the FUJI II and FUJI IX groups are shown in Table 2. For CMW and S430 groups, there was no bonding characteristic feature evident in the load– displacement curves.
4. Discussion
Fig. 4. The load–displacement diagram from Charalambides’ test shows a steady crack extension by unload–reload load–displacement curve. It also shows asymmetrical crack extension by having double plateau portions.
Recently, the evaluation of the interfacial fracture toughness by Charalambides’ four-point bend delamination test has been used successfully in porcelain– metallic bonding [23–25]. The test seems to evaluate the bonding between cortical bone and cement better than the bilayer compact sandwich test used by Wang and Agrawal [8] because the mode of failure is similar to one of the modes in the clinical situation. In bilayer compact tension tests, although the mixed mode loading occurs at the crack tip interface because of the mismatch of the material’s modulus, the phase angle is still low (approximately 11–171) [11] compared with the bending test. In Charalambides’ four-point bend delamination test, for beams with one relatively low modulus material (E2 =E1 > 3 for bone/PMMA=20/2.8 [11] >3), the phase angle c deviates very little from 451 [15].
Fig. 5. Micrograph of the failure surface of S430-bone bimaterial reveals parts of cement attached on bone surface (cohesive failure).
P. Lucksanasombool et al. / Biomaterials 24 (2003) 1159–1166
1164 Table 2 Properties of bovine bone and cements
Bovine Bone CMW FUJI II FUJI IX S430
Modulus (GPa)
Kc (MPam1/2)
Gss with bovine bone (J/m2)
16.7273.25 2.63 6.79 6.9571.92 2.0970.24
2.3070.27 (10) 1.6570.13 0.18–0.51 0.4270.07 1.0270.09
Cannot be measured 52.9878.50(38.71–59.83)a 44.17711.63(30.41–64.94)b Cannot be measured
Poisson’s ration (n) used in this study=0.3. a Average from 5 successful specimens (Total 11). b Average from 10 successful specimens (Total 11).
Although there have been no publications to directly compare the interfacial fracture toughness between cortical bone and PMMA with this study, there have been publications [7,26,27] on the bonding behaviour of PMMA to bone. Mann et al. [26] reported that the tensile bonding strength of the cement-bone interface depended on the amount of bone interdigitated with PMMA cement. Furthermore, Wang and Agrawal [8] have studied the interfacial fracture toughness of cortical bone-PMMA interface using the bilayer compact sandwich test. However the phase angles of both tests and the bond surface preparation are different for this test. In their study, Wang and Agrawal [8] reported the strain energy release rate measured ranged from 37 to 51 J/m2 whereas in our study, there was no evidence of bonding. The differences appears in large part to result from the roughness of the bony surface, which provide interlocking (physical) between cortical bone and cement as shown in the studies of the metal-PMMA interface [9,10]. In the study by Ohashi et al. [10], the test using the double-cantilever beam (DCB) specimen with the phase angle of 3.51 resulted in the Gss from 3.6 to more than 208 J/m2 with the fine polish to fiber mesh surfaces, respectively. There are two possible explanations for the results of this test if we consider that the bonding between cortical bone and cement results only from mechanical interlocking. One is that our surface finishing could eliminate the interlocking of the cement to bone or this test is less sensitive than Ohashi’s test. However, with the same conditions, the specimens prepared from the conventional glass ionomer cements exhibited significant bonding ability. The interfacial fracture toughness measured was 24–52 J/m2 compared with no bonding at all for PMMA specimens with the same kind of test and the same surface finish. These results are consistent with evidence of bonding of glass ionomer cements to hydroxyapatite within enamel and dentin as mentioned above [28–30] and in addition the bonding of these cements to collagen has been reported by Akinmade [31]. Nevertheless the absence of bonding from the load– displacement curve for the resin-modified glass ionomer cement (S430) despite the anticipated bonding of both
GIC and also the resin component (HEMA) is surprising. The bonding ability of the resin to enamel or dentine relies on the infiltration of the monomer into demineralized dentin or etched enamel, the monomer then polymerizes to form a micro-mechanical bond [28– 30]. Therefore, with bone, with its connective tissue microstructure, the bonding ability of S430 was expected to be similar to that of enamel and dentin. In this study, we found microscopic evidence for bonding of the S430 onto the bone surface, despite the absence of any suggestion of bonding shown in the load–displacement curve. A possible explanation for such contradictory behaviour maybe the residual stresses developed in the S430. The residual stress that develops within the glass ionomer system is due to the setting reaction shrinkage and the hygroscopic expansion after exposure to water, respectively [30,32–38]. The shrinkage stress of the conventional glass ionomer cements is lower than that of the composite resins and the resin-modified glass ionomer cements. The lower shrinkage stress in conventional glass ionomer cements is due largely to the rubbery hydrogel matrix of the setting cement, which expands on hydration reducing the shrinkage stress whereas, in resin-modified glass ionomer cements, the domination of the polymerization reaction over the acid-base reaction causes no or less rubbery structure to reduce the shrinkage stress. For the conventional glass ionomer cement, the shrinkage stress decreases immediately after contacting with water and then increases again because the shrinkage due to the reaction is predominant. In the case of the resin-modified glass ionomer cement, hygroscopic expansion occurs and it counterbalances the shrinkage approximately 5 h after exposure to water. After that, hygroscopic expansion continues causing compressive stress within the GIC [30,37,38]. From our study with the resin-modified glass ionomer cement, S430, HEMA, which is used to modify the conventional glass ionomer cement, is a hydrophilic resin. It expands almost continuously to a considerable extent after exposure to water. Volumetric expansion in water of resin-modified GICs ranges from 3.4% to 11.3% after 24 h [37,38]. In addition from our study, the
P. Lucksanasombool et al. / Biomaterials 24 (2003) 1159–1166
1165
evidence of significant bonding. The resin-modified glass ionomer cement, however, detached from the bovine bone after a period of time due to the stresses from hygroscopic expansion. References
Fig. 6. The dimension of cements plotted with time from one day to 3 months shows continuous expansion of S430 cement up to 3 months whereas no obvious change is observed from Fuji IX cement.
resin-modified GIC (S430) continued to expand (0.63%) up to 3 months (Fig. 6). The resultant stresses are likely to be sufficient to detach the cement from the bone after it had already bonded in the specimen used in this study. The microscopic evidence from the failed bone-S430 interface, suggests fracture resulted from cohesive failure as also occurs for conventional glass ionomer cements. A problem occasionally encountered by us with this four-point bend delamination test was an asymmetrical crack extension, which made the results difficult to interpret. Asymmetric crack advance may result from both variations in Gss with position on both sides of the notch tip and from distortion of the test specimen [19]. Distortion of the test specimen, which occurs within the stiff specimen, might not be the problem in the specimen made from cortical bone and cement because it is able to adjust itself due to the viscoelastic property of bone. However, this problem may be resolved by the use of a floating platen to distribute the load equally to both loading points [19]. For the variations in Gss with position within the specimen, this intrinsic problem is unavoidable with a heterogenous material such as bone. However, it can be solved by modifying Charalambides’ test as suggested by Hofinger et al. [39]. Their modification restricts the crack extension to only one side eliminating the asymmetrical crack extension in the original Charalambides’ test.
5. Conclusions The four-point bend delamination test by Charalambides et al. [15,16] has been applied to evaluate the bonding nature of various cements. The conventional bone cement (CMW) revealed minimal bonding as polishing of the bone eliminated bonding by mechanical interlocking. The glass ionomer cement groups revealed
[1] Jasty M, Maloney W, Bragdon CR, O’Connor D, Haire T, Harris WH. The initiation of failure in cemented femoral components of hip arthroplasties. J Bone Jt Surg Br 1991;73-B:551–8. [2] Verdonschot N, Huiskes N. The effects of cement-stem debonding in THA on the long-term failure probability of cement. J Biomech 1997;38:795–802. [3] Krause WR, Krug W, Miller J. Strength of the cement-bone interface. Clin Orthop Relat Res 1982;163:290–9. [4] Kusleika R, Stupp SI. Mechanical strength of poly(methyl methacrylate) cement-human bone interfaces. J Biomed Mater Res 1983;17:441–58. [5] Askew MJ, Steege JW, Lewis JL, Ranieri JR, Wixson RL. Effect of cement pressure and bone strength on polymethylmethacrylate fixation. J Orthop Res 1984;1:412–20. [6] MacDonald W, Swarts E, Beaver R. Penetration and shear strength of cement-bone interfaces in vivo. Clin Orthop Relat Res 1993;286:283–8. [7] Mann KA, Ayers DC, Werner FW, Nicoletta RJ, Fortino MD. Tensile strength of the cement-bone interface depends on the amount of bone interdigitated with PMMA cement. J Biomech 1997;30:339–46. [8] Wang X, Agrawal CM. Interfacial fracture toughness of tissuebiomaterial systems. J Biomed Mater Res (Applied Biomaterials) 1997;38:1–10. [9] Mann KA, Edidin AA, Ordway NR, Manley MT. Fracture toughness of CoCr alloy-PMMA cement interface. J Biomed Mater Res (Applied Biomaterials) 1997;38:211–9. [10] Ohashi KL, Romero AC, McGowan PD, Maloney WJ, Dauskardt RH. Adhesion and reliability of interfaces in cemented total joint arthroplasties. J Orthop Res 1998;16:705–14. [11] Wang X, Subramanian A, Dhanda R, Agrawal M. Testing of bone-biomaterial interfacial bonding strength: a comparison of different techniques. J Biomed Mater Res (Applied Biomaterials) 1996;33:133–8. [12] Cao HC, Evans AG. An Experimental study of the fracture resistance of bimaterial interfaces. Mech Mater 1989;7:295–304. [13] Evans AG, Hutchinson JW. Effects of non-planarity on the mixed mode fracture resistance of bimaterial interfaces. Acta Metal 1989;37:909–16. [14] Matos PPL, McMeeking RM, Charalambides PG, Drory MD. A method for calculating stress intensities in bimaterial fracture. Int J Fract 1989;40:235–54. [15] Charalambides PG, Lund J, McMeeking RM. A test specimen for determining the fracture resistance of bimaterial interfaces. J Appl Mech 1989;56:77–83. [16] Charalambides PG, Cao HC, Lund J, Evans AG. Development of a test method for measuring the mixed mide fracture resistance of bimaterial interfaces. Mech Mater 1990;8:269–83. [17] Evans AG, Ruhle M, Dalgleish BJ, Charalambides PG. The fracture energy of bimaterial interfaces. Mater Sci Eng 1990;A126:53–64. [18] Varias AG, O’Dowd NP. Failure of bimaterial interfaces. Mater Sci Eng 1990;A126:65–93. [19] Howard SJ, Phillipps AJ, Clyne TW. The interpretation of data from the four-point bend delamination test to measure interfacial fracture toughness. Composites 1993;24:103–12. [20] Rice JR. Elastic fracture mechanics concepts for interfacial cracks. J Appl Mech 1988;55:98–103.
1166
P. Lucksanasombool et al. / Biomaterials 24 (2003) 1159–1166
[21] Lipson SF, Katz JL. The relationship between elastic properties and microstructure of bovine cortical bone. J Biomech 1984;17: 231–40. [22] Martin RB, Boardman DL. The effects of collagen fiber orientation, porosity, density, and mineralization on bovine cortical bone bending properties. J Biomech 1993;26:1047–54. [23] Chung HG, Swain MV, Mori T. Strain energy release rate of titanium-porcelain interfacial bonding. Biomaterials 1997;18: 1553–7. [24] Suansuwan N, Swain MV. The adhesion of dental porcelain to various metals. J Aust Ceram Soc 1998;34:189–94. [25] Suansuwan N, Swain MV. New approach for evaluating metalporcelain interfacial bonding. Int J Prosthodont 1999;12:547–52. [26] Mann KA, Werner FW, Ayers DC. Modeling the tensile behavior of the cement-bone interface using nonlinear fracture mechanics. J Biomech Eng 1997;119:175–8. [27] Skripitz R, Aspenberg P. Attachment of PMMA cement to bone: force measurements in rats. Biomaterials 1999;20:351–6. [28] Wilson AD, McLean JW. Glass-ionomer cement. Chicago, IL: Quintessence Publishing Co., Inc., 1988. [29] Hunt PR. Glass ionomer: the next generation. Proceedings of the Second International Symposium on Glass Ionomers. International Symposia in Dentistry, PC, Philadelphia, PA, 1994. . IA. Advances in glass-ionomer cements. [30] Davidson CL, Mjor Illinois: Quintessence Publishing Co Inc, 1999.
[31] Akinmade A. The adhesion of glass polyalkenoate cements to collagen. J Dent Res 1994;73:181. [32] Grajower R, Guelmann M. Dimensional changes during setting of a glass ionomer filling material. Quint Int 1989;20:505–11. [33] Kanchanavasita W, Pearson GJ, Anstice HM. Temperature rise in ion-leachable cements during setting reaction. Biomaterials 1995;16:1261–5. [34] Kanchanavasita W, Pearson GJ, Anstice HM. Influence of humidity on dimensional stability of a range of ion-leachable cements. Biomaterials 1995;16:921–9. [35] Mitchell CA, Orr JF, Kennedy JG. Factors Influencing the failure of dental glass ionomer luting cement due to contraction. Biomaterials 1995;16:11–6. [36] Sidhu SK, Sherriff M, Watson TF. The effects of maturity and dehydration shrinkage on resin-modified glass-ionomer restorations. J Dent Res 1997;76:1495–501. [37] Cattani-Lorente MA, Dupuis V, Payan J, Moya F, Meyer JM. Effect of water on the physical properties of resin-modified glass ionomer cements. Dent Mater 1999;15:71–8. [38] Watts DC, Kisumbi BK, Toworfe GK. Dimensional changes of resin/ionomer restoratives in aqueous and neutral media. Dent Mater 2000;16:89–96. [39] Hofinger I, Oechsner M, Bahr H-A, Swain MV. Modified fourpoint bending specimen for determining the interface fracture energy for thin, brittle layers. Int J Fract 1998;92:213–20.
Interfacial fracture toughness between bovine cortical bone and cements P. Lucksanasomboola,c, W.A.J Higgsb,c, R.J.E.D Higgsa,b, M.V Swainb,c,* a Faculty of Medicine, University of Sydney, Australia Department of Mechanical & Mechatronic Engineering, University of Sydney, Australia c Biomaterials Science Research Unit, Faculty of Dentistry, University of Sydney, Suite G11, National Innovation Centre, Australian Technology Park, Eveleigh, NSW 1430, Australia b
Received 10 July 2001; accepted 20 July 2002
Abstract To evaluate the bonding strength of the interfaces within the cemented arthroplasty system, various mechanical tests have been used. Conventional push-out and pull-out tests cannot reveal the actual bonding property of the interface because of the significant influence of surface roughness on the measured adhesion and the failure to account for the mismatch of elastic modulus across the interface. An alternative fracture mechanics approach, which considers the mix of opening and shear modes of the crack tip loading associated with the testing system and the elastic mismatch of materials across the interface, was used to evaluate the bonding ability of various cements. The four-point bend interfacial delamination test by Charalambides et al. (J. Appl. Mech. 56 (1989) 77; Mech. Mater. 8 (1990) 269) was used to quantify the bonding ability of cements. This method is arguably more suitable since the applied loading mode is comparable to the nature of loading within the prosthetic system, which is primarily bending. The bovine bone specimens were polished to mirror finish to eliminate bonding by mechanical interlocking. The results revealed minimal bonding for the conventional bone cement (PMMA) whereas substantial bonding was evident for the glass-ionomer cements tested. However, only the conventional glass-ionomer cements showed evidence of bonding on testing, while the resin-modified glass-ionomer cement (poly-HEMA) did not. The latter appeared to debond before testing because of excessive expansion stresses associated with swelling in water. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Interfacial fracture toughness; Critical strain energy release rate; Phase angle; Bone cements; Glass ionomer cements
1. Introduction Debonding of the cement-metal interface has been implicated as a major site of failure initiation [1,2] in total joint replacement. There have been many attempts to improve the efficacy of bonding by such means as improved cementing techniques, and the addition of bonding agents to existing cement systems. In order to evaluate the improvement in bonding, conventional push-out and pull-out tests [3–7] and fracture mechanics approaches [8–10] have been used. Prior investigations studied the bonding between bone and bone cement using conventional strength methods such as push-out and pull-out tests [3–7]. The resultant *Corresponding author. Tel.: +61-2-9351-1814; fax: +61-2-93511815. E-mail address: [email protected] (M.V. Swain).
failure characteristics were unable to reveal an actual bonding property of the interface and inter-comparison has proved exceptionally difficult because of uncertain types of failure (bone fracture, cement fracture and/or mixed) (Table 1). Various types of failures occurred in push-out and pull-out tests because of the mix of loading modes which depend on the loading geometry and the mismatch of the elastic modulus of the two materials. Wang et al. [11] conducted bimaterial bonding strength tests using different techniques and found that push-out and pull-out tests were sensitive to specimen configuration and had less reproducibility due to the material mismatch across the interface. This mismatch resulted in the complex stress distribution along the interface. In subsequent studies, a fracture mechanics approach was introduced with the aim of better evaluating debonding at the interface. Tests of this type are more colloquially referred to as interfacial
0142-9612/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 1 4 2 - 9 6 1 2 ( 0 2 ) 0 0 4 6 4 - 7
P. Lucksanasombool et al. / Biomaterials 24 (2003) 1159–1166
1160 Table 1 Previous interfacial strength studies Studies
Test methods
Failure sites
Interfacial strength
Krause et al. (1982)
Tensile pull-out test
2.7271.42 MPa
Krause et al. (1982)
Shear-cylindrical specimens test
Through cancellous-cement interface (mo cleaning; fingerpacked) Through cement (high-intensity lavage; pressure injected) Not be described (no cleaning; finger-packed)
Kusleika and Stupp (1983)
Tensile pull-out test
Askew et al. (1984)
Tensile test
Macdonald et al. (1993) Mann et al. (1997)
Shear pull-out test Tensile pull-out test
Not be described (high-intensity lavage; pressure injected) Cement fracture (most common) Mixed fracture Bone fracture (less common) No penetration Penetration cement was pulled out from bone Bone fracture Cement fracture Not be described Through cancellous-cement interface (40 specimens) Through composite (27 specimens) Through bone adjacent to the interface (4 specimens)
fracture toughness tests [12–19]. This conditioned approach is better able to reveal the interfacial bonding as it considers both the loading mode and the material mismatch within the testing procedures [11]. Using a fracture mechanics approach, the mixture of materials across the interface and the testing procedure result in a mixed opening and shear loading mode developing at the crack tip that is characterized by the phase angle, c; which is equal to the ratio of the shearing (mode II) to opening (mode I) stress intensity factors. Therefore, in pure mode I, the phase angle is 01 and is 901 in pure mode II. With differing specimen geometry and applied loading, the resultant interfacial toughness varies because the phase angle will change (Fig. 1). The resultant interfacial fracture toughness can be described in terms of the stress intensity factor, K; or the critical strain energy release rate, G: However, the bimaterial stress intensity factor and phase angle c are scale sensitive [20]. Therefore, it is more convenient for comparison to consider the interface fracture resistance locus in terms of trends in critical strain energy release rate G; which is scale insensitive [15]. The four-point bend delamination test (Fig. 1), which produces stable crack propagation, was developed for measuring the fracture resistance of bimaterial interfaces by Charalambides et al. [15,16]. The test is appropriate for testing the interfaces between cortical bone and bone cement because the loading mode is relevant to one of the loading modes (bending) within the prosthetic system. In this study, the four-point bend delamination test was used to measure the bonding ability between bovine cortical bone and three different types of cements, consisting of the conventional bone cement (PMMA), two conventional glass ionomer cements
8.5074.28 MPa 9.5574.06 MPa 41.9975.89 MPa 6–9 MPa 5–8 MPa 4–6 MPa Cannot be determined
3.2–7.3 MPa 0.9870.69 MPa 1.6870.66 MPa 1.4970.7 MPa
(FUJI II & IX), and the novel resin-modified glass ionomer cement (S430). 2. Materials and methods 2.1. Four-point bend delamination test: theoretical framework The specimen consists of a notched, bimaterial flexural beam (Fig. 1). A central notch exits through the thickness of one material and a symmetrical precrack is induced along the interface. When the specimen is loaded in four-point bending, a steady-state scheme can be obtained within the inner loading lines [15] and at least when the interface crack length significantly exceeds the thickness of the upper layer of the beam. The steady-state value, Gss ; can be determined analytically by simply comparing the difference in the strain energy in the uncracked and cracked beam. From Euler–Bernoulli beam theory and plain strain conditions, the strain energy per unit cross-section, U; can be expressed in terms of the applied moment M as ð1 n2 ÞM 2 ; 2EI where I is the second moment of area per unit width, E is the elastic modulus of the beam, and n is the Possion’s ratio. When the equation is applied to the bimaterial for the steady-state value, Gss ; which is the difference in the strain energy in the uncracked and cracked beam, the equation is M 2 ð1 n22 Þ 1 l Gss ¼ ; 2E2 I2 Ic U¼
P. Lucksanasombool et al. / Biomaterials 24 (2003) 1159–1166
1161
Fig. 1. Schematic illustration of various specimen configurations with their phase angles.
where the subscript 2 describes the properties of the unnotched layer, the subscript c refers the property of the composite beam and l is l¼
E2 ð1 n21 Þ : E1 ð1 n22 Þ
where Z is equal ( 3 1 Z¼ 2 ðh2 =hÞ3 "
l 3 3 ðh1 =hÞ þ lðh2 =hÞ þ 3lðh1 h2 =h2 Þðh1 =h þ lh2 =hÞ1
The subscript 1 describes the properties of the notched layer of composite beam. Ic can be calculated as the equation: Ic ¼ h31 =12 þ lh32 12 þ lh1 h2 ðh1 þ h2 Þ2 =4ðh1 þ lh2 Þ; where hn is the height or thickness of each layer and I2 can be expressed as
#) :
The steady-state strain energy release rate, Gss ; with the unit of J/m2 has also been analysed in terms of the effects of the thickness and moduli of each material constructing the bimaterial specimen, the frictional effects, the residual stress effects, and also the interpretation of the load–displacement curves [15,16,19].
I2 ¼ h32 =12:
2.2. Specimen preparation and testing
Noting that the moment per unit width M ¼ Pl=2b; with P being the total load, b is the width of the beam, and l is the space between inner and outer loading lines, the strain energy release rate turns into:
Whole bovine tibias of unknown age and post mortem period were used. The bovine bones were soaked with normal saline and wrapped with normal saline soaked gauze. They were kept in a plastic bag and stored at 201C before processing. In order to sustain less variability of the bone specimens [21,22], cortical bone mainly from the posterior aspect of bovine tibias was cut and milled to rectangular bars of size 2.5 10.0 25.0 mm3. The longitudinal axis of the bars was parallel to the longitudinal axis of the bone. The intramedullary aspect of the specimens, which was bonded with the cement, was sequential polished with finer diamond powders until a mirror finish was attained. The bone specimens were maintained wet in saline during all stages of tissue preparation.
E2 Gss h3 b2 =P2 l 2 ð1 n22 Þ ¼ ð3=2Þf1=ðh2 =hÞ3 l=fðh1 =hÞ3 þ lðh2 =hÞ3 þ 3lðh1 h2 =h2 Þ 1 h1 =h þ lh2 =h gg: When the constant value in the equation is replaced with Z, the equation for the steady-state strain energy release rate, Gss ; can be rewritten as Gss ¼
ZP2 l 2 ð1 n22 Þ ; E 2 b2 h3
1162
P. Lucksanasombool et al. / Biomaterials 24 (2003) 1159–1166
Cements used in this study were a conventional bone cement, PMMA (CMW), a conventional glass ionomer cement (FUJI II), a high viscosity conventional glass ionomer cement (FUJI IX), and a resin-modified glass ionomer cement (S430). Cements were mixed manually with the powder/liquid ratio and mixing instructions according to each manufacturer’s recommendations. To produce the bimaterial specimens, the cement was packed into the 5.0 10.0 25.0 mm3 mould over the sponge dried bone bar and pressurized until the cement set. Eleven bimaterial specimens for each cement group were made. The bimaterial specimens were cleaned and ground to remove the excessive cement then they were kept at 371C normal saline for 1 week. The bimaterial specimens were notched through the cement layer using a low-speed diamond blade saw and precracked in a narrow (o5 mm) three-point bending jig, together with simultaneous observation using an optical microscope, to limit the extension of the precracks before testing. The specimens were loaded in four-point bending with the loading rate equal to 0.1 mm/min using a Shimadzu testing machine. An optical microscope was used to observe crack front extension on both sides of the introduced notch. Observation of the fracture surfaces was performed using a scanning electron microscope (SEM). A thin carbon coating was applied to the fracture surfaces of the specimens to minimize electrostatic charging. Some bovine bone specimens as well as cement bars fabricated from the same mould were also tested for the
moduli in the same storage and testing conditions as the bimaterial.
3. Results The observed crack extension from the initial notch revealed two typical forms, symmetrical and asymmetrical crack extensions (Fig. 2). For the bimaterial specimens prepared from FUJI II and IX, the observed crack extension phenomena can be related well with the loading. In symmetrical crack extension, the load increased linearly until the crack initiated whereupon the load remained almost constant as indicated by the plateau portion of the load–displacement curve, which showed less inclination than the other portions. The load increased linearly again after the crack propagated beyond the inner loading points (Fig. 3). The crack extension in this case was steady as can be seen on the unload–reload curve (Fig. 4). In case of asymmetrical crack extension, the loading behaviour was the same, but having two shorter plateau portions corresponding with the crack extension on each side (Fig. 4). For the specimens prepared from CMW and S430, the crack extension occurred immediately after loading in some specimens, whereas, in the others, no indication of crack extension was evident from the force–displacement curves despite application of significantly high loads and displacements. In the specimens, which displayed no obvious crack extension during loading,
Fig. 2. Micrographs of crack extension in Charalambides’ test: (a) symmetrical crack extension, (b) asymmetrical crack extension.
P. Lucksanasombool et al. / Biomaterials 24 (2003) 1159–1166
300 CMW-Bone FUJI II-Bone FUJI IX-Bone S430-Bone
250
Load (N )
200 150 100 50 0 0
0.1
0.2 Displacement (mm.)
0.3
0.4
Fig. 3. The load–displacement diagram from Charalambides’ test of various cement-bone bimaterials. The plateau portion (Fuji II and Fuji IX bimaterials) reveals interfacial bonding.
1163
it was found that both layers were completely detached from each other. However, upon SEM observation, the bone surface of S430 specimens revealed cohesive failure (Fig. 5) of cement material similar to other GICs whereas there was no cement attached to the bone for CMW specimens. For the calculation of the strain energy release rate, the plateau loads, which were determined by the obvious changes of the inclination of the load–displacement, were averaged. The calculated strain energy release rate, Gss ; of the FUJI II and FUJI IX groups are shown in Table 2. For CMW and S430 groups, there was no bonding characteristic feature evident in the load– displacement curves.
4. Discussion
Fig. 4. The load–displacement diagram from Charalambides’ test shows a steady crack extension by unload–reload load–displacement curve. It also shows asymmetrical crack extension by having double plateau portions.
Recently, the evaluation of the interfacial fracture toughness by Charalambides’ four-point bend delamination test has been used successfully in porcelain– metallic bonding [23–25]. The test seems to evaluate the bonding between cortical bone and cement better than the bilayer compact sandwich test used by Wang and Agrawal [8] because the mode of failure is similar to one of the modes in the clinical situation. In bilayer compact tension tests, although the mixed mode loading occurs at the crack tip interface because of the mismatch of the material’s modulus, the phase angle is still low (approximately 11–171) [11] compared with the bending test. In Charalambides’ four-point bend delamination test, for beams with one relatively low modulus material (E2 =E1 > 3 for bone/PMMA=20/2.8 [11] >3), the phase angle c deviates very little from 451 [15].
Fig. 5. Micrograph of the failure surface of S430-bone bimaterial reveals parts of cement attached on bone surface (cohesive failure).
P. Lucksanasombool et al. / Biomaterials 24 (2003) 1159–1166
1164 Table 2 Properties of bovine bone and cements
Bovine Bone CMW FUJI II FUJI IX S430
Modulus (GPa)
Kc (MPam1/2)
Gss with bovine bone (J/m2)
16.7273.25 2.63 6.79 6.9571.92 2.0970.24
2.3070.27 (10) 1.6570.13 0.18–0.51 0.4270.07 1.0270.09
Cannot be measured 52.9878.50(38.71–59.83)a 44.17711.63(30.41–64.94)b Cannot be measured
Poisson’s ration (n) used in this study=0.3. a Average from 5 successful specimens (Total 11). b Average from 10 successful specimens (Total 11).
Although there have been no publications to directly compare the interfacial fracture toughness between cortical bone and PMMA with this study, there have been publications [7,26,27] on the bonding behaviour of PMMA to bone. Mann et al. [26] reported that the tensile bonding strength of the cement-bone interface depended on the amount of bone interdigitated with PMMA cement. Furthermore, Wang and Agrawal [8] have studied the interfacial fracture toughness of cortical bone-PMMA interface using the bilayer compact sandwich test. However the phase angles of both tests and the bond surface preparation are different for this test. In their study, Wang and Agrawal [8] reported the strain energy release rate measured ranged from 37 to 51 J/m2 whereas in our study, there was no evidence of bonding. The differences appears in large part to result from the roughness of the bony surface, which provide interlocking (physical) between cortical bone and cement as shown in the studies of the metal-PMMA interface [9,10]. In the study by Ohashi et al. [10], the test using the double-cantilever beam (DCB) specimen with the phase angle of 3.51 resulted in the Gss from 3.6 to more than 208 J/m2 with the fine polish to fiber mesh surfaces, respectively. There are two possible explanations for the results of this test if we consider that the bonding between cortical bone and cement results only from mechanical interlocking. One is that our surface finishing could eliminate the interlocking of the cement to bone or this test is less sensitive than Ohashi’s test. However, with the same conditions, the specimens prepared from the conventional glass ionomer cements exhibited significant bonding ability. The interfacial fracture toughness measured was 24–52 J/m2 compared with no bonding at all for PMMA specimens with the same kind of test and the same surface finish. These results are consistent with evidence of bonding of glass ionomer cements to hydroxyapatite within enamel and dentin as mentioned above [28–30] and in addition the bonding of these cements to collagen has been reported by Akinmade [31]. Nevertheless the absence of bonding from the load– displacement curve for the resin-modified glass ionomer cement (S430) despite the anticipated bonding of both
GIC and also the resin component (HEMA) is surprising. The bonding ability of the resin to enamel or dentine relies on the infiltration of the monomer into demineralized dentin or etched enamel, the monomer then polymerizes to form a micro-mechanical bond [28– 30]. Therefore, with bone, with its connective tissue microstructure, the bonding ability of S430 was expected to be similar to that of enamel and dentin. In this study, we found microscopic evidence for bonding of the S430 onto the bone surface, despite the absence of any suggestion of bonding shown in the load–displacement curve. A possible explanation for such contradictory behaviour maybe the residual stresses developed in the S430. The residual stress that develops within the glass ionomer system is due to the setting reaction shrinkage and the hygroscopic expansion after exposure to water, respectively [30,32–38]. The shrinkage stress of the conventional glass ionomer cements is lower than that of the composite resins and the resin-modified glass ionomer cements. The lower shrinkage stress in conventional glass ionomer cements is due largely to the rubbery hydrogel matrix of the setting cement, which expands on hydration reducing the shrinkage stress whereas, in resin-modified glass ionomer cements, the domination of the polymerization reaction over the acid-base reaction causes no or less rubbery structure to reduce the shrinkage stress. For the conventional glass ionomer cement, the shrinkage stress decreases immediately after contacting with water and then increases again because the shrinkage due to the reaction is predominant. In the case of the resin-modified glass ionomer cement, hygroscopic expansion occurs and it counterbalances the shrinkage approximately 5 h after exposure to water. After that, hygroscopic expansion continues causing compressive stress within the GIC [30,37,38]. From our study with the resin-modified glass ionomer cement, S430, HEMA, which is used to modify the conventional glass ionomer cement, is a hydrophilic resin. It expands almost continuously to a considerable extent after exposure to water. Volumetric expansion in water of resin-modified GICs ranges from 3.4% to 11.3% after 24 h [37,38]. In addition from our study, the
P. Lucksanasombool et al. / Biomaterials 24 (2003) 1159–1166
1165
evidence of significant bonding. The resin-modified glass ionomer cement, however, detached from the bovine bone after a period of time due to the stresses from hygroscopic expansion. References
Fig. 6. The dimension of cements plotted with time from one day to 3 months shows continuous expansion of S430 cement up to 3 months whereas no obvious change is observed from Fuji IX cement.
resin-modified GIC (S430) continued to expand (0.63%) up to 3 months (Fig. 6). The resultant stresses are likely to be sufficient to detach the cement from the bone after it had already bonded in the specimen used in this study. The microscopic evidence from the failed bone-S430 interface, suggests fracture resulted from cohesive failure as also occurs for conventional glass ionomer cements. A problem occasionally encountered by us with this four-point bend delamination test was an asymmetrical crack extension, which made the results difficult to interpret. Asymmetric crack advance may result from both variations in Gss with position on both sides of the notch tip and from distortion of the test specimen [19]. Distortion of the test specimen, which occurs within the stiff specimen, might not be the problem in the specimen made from cortical bone and cement because it is able to adjust itself due to the viscoelastic property of bone. However, this problem may be resolved by the use of a floating platen to distribute the load equally to both loading points [19]. For the variations in Gss with position within the specimen, this intrinsic problem is unavoidable with a heterogenous material such as bone. However, it can be solved by modifying Charalambides’ test as suggested by Hofinger et al. [39]. Their modification restricts the crack extension to only one side eliminating the asymmetrical crack extension in the original Charalambides’ test.
5. Conclusions The four-point bend delamination test by Charalambides et al. [15,16] has been applied to evaluate the bonding nature of various cements. The conventional bone cement (CMW) revealed minimal bonding as polishing of the bone eliminated bonding by mechanical interlocking. The glass ionomer cement groups revealed
[1] Jasty M, Maloney W, Bragdon CR, O’Connor D, Haire T, Harris WH. The initiation of failure in cemented femoral components of hip arthroplasties. J Bone Jt Surg Br 1991;73-B:551–8. [2] Verdonschot N, Huiskes N. The effects of cement-stem debonding in THA on the long-term failure probability of cement. J Biomech 1997;38:795–802. [3] Krause WR, Krug W, Miller J. Strength of the cement-bone interface. Clin Orthop Relat Res 1982;163:290–9. [4] Kusleika R, Stupp SI. Mechanical strength of poly(methyl methacrylate) cement-human bone interfaces. J Biomed Mater Res 1983;17:441–58. [5] Askew MJ, Steege JW, Lewis JL, Ranieri JR, Wixson RL. Effect of cement pressure and bone strength on polymethylmethacrylate fixation. J Orthop Res 1984;1:412–20. [6] MacDonald W, Swarts E, Beaver R. Penetration and shear strength of cement-bone interfaces in vivo. Clin Orthop Relat Res 1993;286:283–8. [7] Mann KA, Ayers DC, Werner FW, Nicoletta RJ, Fortino MD. Tensile strength of the cement-bone interface depends on the amount of bone interdigitated with PMMA cement. J Biomech 1997;30:339–46. [8] Wang X, Agrawal CM. Interfacial fracture toughness of tissuebiomaterial systems. J Biomed Mater Res (Applied Biomaterials) 1997;38:1–10. [9] Mann KA, Edidin AA, Ordway NR, Manley MT. Fracture toughness of CoCr alloy-PMMA cement interface. J Biomed Mater Res (Applied Biomaterials) 1997;38:211–9. [10] Ohashi KL, Romero AC, McGowan PD, Maloney WJ, Dauskardt RH. Adhesion and reliability of interfaces in cemented total joint arthroplasties. J Orthop Res 1998;16:705–14. [11] Wang X, Subramanian A, Dhanda R, Agrawal M. Testing of bone-biomaterial interfacial bonding strength: a comparison of different techniques. J Biomed Mater Res (Applied Biomaterials) 1996;33:133–8. [12] Cao HC, Evans AG. An Experimental study of the fracture resistance of bimaterial interfaces. Mech Mater 1989;7:295–304. [13] Evans AG, Hutchinson JW. Effects of non-planarity on the mixed mode fracture resistance of bimaterial interfaces. Acta Metal 1989;37:909–16. [14] Matos PPL, McMeeking RM, Charalambides PG, Drory MD. A method for calculating stress intensities in bimaterial fracture. Int J Fract 1989;40:235–54. [15] Charalambides PG, Lund J, McMeeking RM. A test specimen for determining the fracture resistance of bimaterial interfaces. J Appl Mech 1989;56:77–83. [16] Charalambides PG, Cao HC, Lund J, Evans AG. Development of a test method for measuring the mixed mide fracture resistance of bimaterial interfaces. Mech Mater 1990;8:269–83. [17] Evans AG, Ruhle M, Dalgleish BJ, Charalambides PG. The fracture energy of bimaterial interfaces. Mater Sci Eng 1990;A126:53–64. [18] Varias AG, O’Dowd NP. Failure of bimaterial interfaces. Mater Sci Eng 1990;A126:65–93. [19] Howard SJ, Phillipps AJ, Clyne TW. The interpretation of data from the four-point bend delamination test to measure interfacial fracture toughness. Composites 1993;24:103–12. [20] Rice JR. Elastic fracture mechanics concepts for interfacial cracks. J Appl Mech 1988;55:98–103.
1166
P. Lucksanasombool et al. / Biomaterials 24 (2003) 1159–1166
[21] Lipson SF, Katz JL. The relationship between elastic properties and microstructure of bovine cortical bone. J Biomech 1984;17: 231–40. [22] Martin RB, Boardman DL. The effects of collagen fiber orientation, porosity, density, and mineralization on bovine cortical bone bending properties. J Biomech 1993;26:1047–54. [23] Chung HG, Swain MV, Mori T. Strain energy release rate of titanium-porcelain interfacial bonding. Biomaterials 1997;18: 1553–7. [24] Suansuwan N, Swain MV. The adhesion of dental porcelain to various metals. J Aust Ceram Soc 1998;34:189–94. [25] Suansuwan N, Swain MV. New approach for evaluating metalporcelain interfacial bonding. Int J Prosthodont 1999;12:547–52. [26] Mann KA, Werner FW, Ayers DC. Modeling the tensile behavior of the cement-bone interface using nonlinear fracture mechanics. J Biomech Eng 1997;119:175–8. [27] Skripitz R, Aspenberg P. Attachment of PMMA cement to bone: force measurements in rats. Biomaterials 1999;20:351–6. [28] Wilson AD, McLean JW. Glass-ionomer cement. Chicago, IL: Quintessence Publishing Co., Inc., 1988. [29] Hunt PR. Glass ionomer: the next generation. Proceedings of the Second International Symposium on Glass Ionomers. International Symposia in Dentistry, PC, Philadelphia, PA, 1994. . IA. Advances in glass-ionomer cements. [30] Davidson CL, Mjor Illinois: Quintessence Publishing Co Inc, 1999.
[31] Akinmade A. The adhesion of glass polyalkenoate cements to collagen. J Dent Res 1994;73:181. [32] Grajower R, Guelmann M. Dimensional changes during setting of a glass ionomer filling material. Quint Int 1989;20:505–11. [33] Kanchanavasita W, Pearson GJ, Anstice HM. Temperature rise in ion-leachable cements during setting reaction. Biomaterials 1995;16:1261–5. [34] Kanchanavasita W, Pearson GJ, Anstice HM. Influence of humidity on dimensional stability of a range of ion-leachable cements. Biomaterials 1995;16:921–9. [35] Mitchell CA, Orr JF, Kennedy JG. Factors Influencing the failure of dental glass ionomer luting cement due to contraction. Biomaterials 1995;16:11–6. [36] Sidhu SK, Sherriff M, Watson TF. The effects of maturity and dehydration shrinkage on resin-modified glass-ionomer restorations. J Dent Res 1997;76:1495–501. [37] Cattani-Lorente MA, Dupuis V, Payan J, Moya F, Meyer JM. Effect of water on the physical properties of resin-modified glass ionomer cements. Dent Mater 1999;15:71–8. [38] Watts DC, Kisumbi BK, Toworfe GK. Dimensional changes of resin/ionomer restoratives in aqueous and neutral media. Dent Mater 2000;16:89–96. [39] Hofinger I, Oechsner M, Bahr H-A, Swain MV. Modified fourpoint bending specimen for determining the interface fracture energy for thin, brittle layers. Int J Fract 1998;92:213–20.