A fractographic analysis of in vivo poly(methy1 methacrylate) bone cement failure mechanisms L. D. Timmie Topoleski University of Pennsylvania, Department of Bioengineering, School of Engineering and Applied Science, Philadelphia, Pennsylvania 19104 Paul Ducheyne University of Pennsylvania, Department of Bioengineering, School of Engineering and Applied Science; and Deparfment of Orthopaedics, School of Medicine, Philadelphia, Pennsylvania 19104 John M. Cuckler University of Pennsylvania, Department of Orthopaedics, School of Medicine, Philadelphia, Pennsylvania 19104 Cementing with poly(methy1 methacrylate) (PMMA) is a common means of fixing total hip prostheses. Bone cement fails mechanically, and subsequent loosening frequently requires correction via revision surgery. An initial step in optimizing bone cement properties is to establish which properties are critical to the material’s in uiuo performance. The objectives were to discern the critical in uiuo failure mechanisms of bone cement. Fracture surfaces of bone cement specimens that failed in uivo were compared with fatigue and rapid fracture surfaces created in uitro. In uivo fracture processes of bone cement were positively identified and explained by the
elucidation of PMMA fracture micromechanisms. The ex viuo fracture surfaces are remarkably similar to in vitro fatigue fracture surfaces. The fractographic data document that the primary in vivo failure mechanism of bone cement is fatigue, and the fatigue cracks grow by developing a microcraze shower damage zone. Agglomerates of BaS04 particles can be implicated in some bone cement failures, large flaws or voids in vivo can lead to a rapid, unstable fracture, pores in the PMMA mass have a clear influence on a propagating crack, and wear of the fracture surfaces occurs, and may produce PMMA debris, exacerbating bone destruction.
INTRODUCTION
Cementing with poly(methy1 methacrylate) (PMMA) is the predominant means of fixing total hip prostheses. The success rate of primary cemented hip arthroplasties at ten years exceeds 90% in patients 60 or more years of age when proper cementing techniques are however, bone cement does fail mechanically. In addition, the incidence of cemented prosthesis failure in patients younger than fifty is considerably higher.44 Failure is often associated with several clinically observed modes of prosthesis loosening.7,8The mechanical failure of bone cement can result further in the production of cement particles. The particles interact with the surrounding Journal of Biomedical Materials Research, Vol. 24, 135-154 (1990) CCC 0021-93O4/90/.020135-20$04.00 0 1990 John Wiley & Sons, Inc.
136
TOPOLESKI, DUCHEYNE, AND CUCKLER
tissues, and may contribute to an inflammatory response, increasing bone destruction, and accelerating prosthesis loosening. Prosthesis loosening frequently requires correction via revision surgery. The improvement of the bone cement integrity is fundamental to the longevity of cemented prosthesis systems, The identification of properties critical to the in vivo performance of bone cement will focus bone cement optimization efforts. Failed bone cement specimens retrieved from patients provide the unique source of evidence for studying in-service failure mechanisms of this material. Fractographic analysis of retrieved failed cement is the most useful method for observing failure patterns and establishing significant mechanical properties. Few studies have reported analyses of bone cement specimens retrieved from in vivo service, Willert et al.'" presented scanning electron microscopy (SEM) data describing the morphology of bone cement retrieved from humans. They examined the endosteal surfaces of the PMMA, and did not consider in vivo fracture surfaces of bone cement. Bone cement fractography has been reported infrequently, primarily as an adjunct to in vitro bone cement material characterization studies. Freitag and Cannon'' identified the extent of fatigue crack growth before catastrophic failure by observing a rough to smooth transition of the fracture surface. A review of other bone cement fractography12-18shows that distinct fracture surface morphologies are associated with the several modes of fracture possible in materials. The mechanisms of bone cement failure in vivo have not been investigated in depth; thus, our objective was to elucidate the in vivo failure mechanisms of bone cement. Fracture surfaces of bone cement specimens that failed in vivo, were compared with fatigue, slow continuous, and rapid continuous fracture surfaces created in vitro. In vivo fracture surface morphologies of bone cement were identified and related to PMMA fracture micromechanisms. METHODS AND MATERIALS
Specimens of PMMA, from 12 failed cemented arthroplasties, were obtained during revision surgery. The cement fractures were generally visible in preoperative radiographs, except in three cases. Clinical data corresponding to each specimen of fractured cement are summarized in Table I. The average age of the patients included in this study was 73 years. The average time between initial implant and revision was 8 years, 9 months. Figure 1(A) illustrates a radiograph showing multiple cement fractures. Loosening and subsequent revision of the illustrated prosthesis (case #7) was performed 3 years after cement fractures were first observed radiographically. Six of the prostheses that were revised and contributed bone cement to this study showed only single, longitudinal bone cement fractures in preoperative radiographs. In contrast to an observation often made in the l i t e r a t ~ r e , 'these ~ , ~ ~ cases were revised for loosening.
PMMA FRACTOGRAPHY
137
TABLE I Clinical Data: In V i m Specimen Sources
Specimen
Age at Revision 63 81 81 71 61
6 7 8 9
84
10
45
11 12
62 69
80 66 73
Sex F F F M M F F F F
M M M
Time between Initial Surgery and Revision 13 years, 6 years, 2 years, 14 years, 2 years,
Time between Cement Failure Diagnosis and Revision
9 months
a.
5 months 11 months 7 months 7 months
1year, 2 months 1 year, 5 months
unknown 7 years, 0 months 14 years, 0 months 2 years, 2 months 14 years, 0 months 2 years, 9 months 6 years, 5 months
0 0
unknown 3 years, 0 months 1 yearb 2 months unknown 2 years 3 years
Initial Implant Type Charnley Charnley Mueller Mueller Bipolar (long stem) T 28 Mueller Mueller Mueller Bipolar Mueller Mueller
aZero (0) time indicates cement failure diagnosed at surgery. bAbsence of specific number of months indicates record document only the year involved.
Only pieces of fractured cement that were removed easily from the medullary canal were included in the study; thus artifacts on the fracture surfaces caused by instrument damage were avoided. All specimens were cleaned in a detergent* to remove biological debris. They were then cut using a slow-speed diamond saw, and sputter-coated with gold. A photograph of a prepared fracture specimen (from case #7) is shown in Figure l(B). A commercially available bone cementt was used to make the in vitro specimens. The monomer was chilled at O°C at least 20 min prior to mixing. The bone cement was mixed according to the instructions of the package insert. Fracture surfaces were created irt vitvo by four distinct methods: 1) Fatigue surfaces, Type I, were produced by fatiguing rectangular prismatic bars (10 X 10 x 60 mm) in three-point bending. A 2.0-mm-long notch was machined at the specimen mid-span with a low-speed diamond saw, and then a sharp groove was cut into the tip of the notch with a scalpel. A fatigue crack approximately 2.5-3.5 mm long was propagated at 10 Hz from the notch tip. Fatiguing was performed under stroke controls at a load that cycled between 22 and 133 N (5-30 lbs.) 2) and 3) Rotating-bending fatigue surfaces, both high (Type 11) and low (Type 111) cycle: Cylindrical waisted beams (5-mm waist diameter) were subjected to fully reversed rotating-bending at a frequency of 30 H Z .The ~ maxi*Liqui-Nox: Alconox, Inc., New York, NY. 'Simplex P: Howmedica, Inc., Rutherford, N.J. $Instron 1331: Instron, Inc., Canton, MA. $RBF-200: Fatigue Dynamics, Inc., Dearborn, MI.
138
TOPOLESKI, DUCHEYNE, AND CUCKLER
(a) Figure 1. (A) Preoperative (revision) radiograph of case #7 (Table I). (B) Photograph of one of the specimens of case #7 prepared for analysis in the SEM. The specimen origin is indicated by the arrow in (A).
mum stresses applied were 9.0 and 17.2 MPa, respectively. The specimens were stored in physiological saline at 37°C for at least 1 week prior to machining, and then replaced into saline at 37°C for at least 1 additional week. 4) Slow continuous fracture surfaces were produced by loading the Type I fatigue specimens at a constant displacement rate of 0.038 mm/s, without fatiguing. The effect of loading the specimen at a slow loading rate, with only the scalpel cut as a starter notch, resulted in initial slow crack growth, followed by an unstable, rapid crack growth. 5) Rapid continuous fracture surfaces were created by loading the Type I specimens at constant displacement rate of 0.076 mm/s, until unstable failure occurred. In view of the studies of Freitag and Cannon11'12 and Beaumont,"13 who documented the effect of test environment on static and fatigue testing in
139
PMMA FRACTOGRAPHY
VJ) Figure 1. (continued)
PMMA bone cement, static tests were performed in air. Fatigue test specimens were maintained constantly moist with physiological saline. RESULTS AND DISCUSSION
The surfaces observed in the majority of the ex vivo analyses were irregular and rough. Example of ex vim surfaces from three separate specimens are included in Figure 2. The morphology of the surfaces resembles irregular stacks of plates or plateaus, and is not peaked or mountainous. Features typical of each of the in vitvo fracture surfaces are shown in Figures 3 and 4. Figure 3(A) shows the overall fracture surface typical of the fracture toughness specimens. The fatigue surface appears as a light colored
140
TOPOLESKI, DUCHEYNE, AND CUCKLER
(4 Figure 2. Ex uiuo fracture surfaces from three different specimens. (A) specimen #6, (B) specimen #7, and (C) specimen #11. Each of the surfaces is irregular and stepped (original magnification 320X, bar = 50 ,urn).
141
PMMA FRACTOGRAPHY
NOTCH
+ t w
FRACTURE
(a)
Figure 3. (A) Overall view of the fracture surface from the fracture toughness preparation. The fatigue surface appears as a light-colored parabolic, or "thumbnail," shaped surface contrasting against the darker, flat, rapid fracture surface. Arrow indicates direction of crack propagation (original magnification 1OX). (B) Enlargement of the fatigue surface. The surface is similar to the ex v i m surfaces shown in Figure 3 . (original magnification 320X, bar = 50 pm). (C) An enlargement of the rapid fracture surface. The twophase nature of bone cement is evident by the clear definition of the beads that make up the powder component of PMMA (original magnification 320X, bar = 50 pm).
parabolic, or "thumbnail", shaped surface contrasting against the darker, flat, rapid fracture surface. Figure 3(B) shows an enlargement of the Type I fatigue surface. Figure 3(C) shows an enlargement of the smooth fracture surface typical of the rapid, unstable, crack growth. The two-phase nature of bone cement is evident by the clear definition of the beads that make up the powder component of PMMA. Figure 4 shows the irregular surface typical of both Type I1 and Type I11 fatigue fracture surfaces. The three types of in vitvo fatigue surfaces are similar. The slow continuous fracture surface was virtually identical to the fatigue surfaces. The similarity of the Type I, 11, and I11 fatigue, and the slow continuous fracture surfaces indicate that the microstructural mechanism causing failure was the same in each case.
142
TOPOLESKI, DUCHEYNE, AND CUCKLER
(c)
Figure 3. (continued)
The ex vivo fracture surfaces are remarkably similar to the in vitro fatigue fracture surfaces. Each of the ex vivo surfaces illustrated in Figure 2 exhibits
PMMA FRACTOGRAPHY
143
Figure 4. Micrograph of the irregular surface associated with Type I1 fatigue cracks (original magnification 320X, bar = 50 Pm). The Type I1 and Type I11 fatigue surfaces are indistinguishable from the Type I fatigue surface (Fig. 3(B)) and the ex uiuo fracture surfaces (Fig. 2).
the stepped surface irregularities that are clear in Figures 3(B) and 4. The proposed crack extension process responsible for the irregular fatigue surface morphology found both ex vivo and in vitra is illustrated by the micrograph presented in Figure 5. The micrograph shows a side view of a three-point bending specimen, looking at the specimen perpendicular to the plane of crack growth. The process illustrated is known as a “microcraze shower,” and occurs when the molecular weight of the PMMA is too low to form fibrils, or craze, across the crack tip damage zone (Fig. 6(A)).22,23 The microcraze shower forms the damage zone ahead of the crack tip (Fig. 6(B)). As energy is added to the system, the microcrazes are connected, and the crack extends (Fig. 6(C)). The characteristic irregular fatigue surface morphology may be the product of the microcraze coalescence. The smooth, rapid fracture surface shown in Figure 3(C) indicates that no microcraze damage zone was formed. The two-phase nature of PMMA bone cement is clear in Figure 3(C). Clawlike markings are observed, but are confined to the fracture path through the polymer beads. The clawlike markings are typical of polymer fracture surfaces where the polymer is of sufficiently high molecular weight, and indicates a layer of crazed The moIecular weight of each phase of the bone cement governs the fracture mechanism, and hence the fracture morphology of the specimens. The material rupture within the mi-
144
TOPOLESKI, DUCHEYNE, AND CUCKLER
Figure 5. Micrograph of the Type I fatigue crack tip. The crack propagation direction is indicated by the black arrow. The microcrazes ahead of the crack tip a r e indicated by t h e white arrow (original magnification 8OX, bar = 100 pm).
crocraze shower probably occurs preferentially through the binding phase PMMA. In addition to the low fatigue crack propagation resistance of the binding phase, the fracture surfaces also evidenced that the prepolymerized phase (the beads) display a relatively low crack propagation resistance, typical for a low molecular weight PMMA.24-26,2s,29 High molecular weight poly(methy1 methacrylate)s, such as industrial PMMA, exhibit fatigue striations. The striations are associated with specific fatigue crack propagation mechanisms. Kim et a1.28showed that fatigue crack propagation continues to decrease with increasing molecular weight even at molecular weights exceeding the maximum molecular weight for which there is a fracture toughness molecular weight dependence. This observation, coupled with the findings of Beaumont and Young,*l suggests that fatigue fracture surface topographies are dependent on crack propagation mechanisms, which, in turn, depend on molecular weight. Beaumont and Youn$ showed that, indeed, the slow fracture surface of high-molecular weight PMMA (viscosity average molecu-
145
PMMA FRACTOGRAPm
A:
6:
+CRACK
__ct)_
DAMAbE
ZONE
(mitrot rare)
C:
Figure 6. A schematic of possible crack propagation mechanisms in PMMA bone cement. (A) If the polymer is of sufficiently high molecular weight, a damage zone will form where fibrils span the crack surfaces. When the stress across the fibrils is great enough, the fibrils will break, and the crack extends. (B) If the molecular weight of the polymer is not high enough to form fibrils across the damage zone, the crack tip may be preceded by a microcraze shower. (C) As the crack opens, the microcrazes coalesce and the crack extends.
lar weight ( M J > 1,000,000) had a smooth appearance, while the slow fracture surface of bone cement (M, = 170,000), was irregular. Neither the ex vivo nor the in vitm fatigue surfaces observed in this study exhibited fatigue striations associated with high molecular weight PMMA. There were several other important features identified through the ex vivo fractography that pertain to the in vivo failure of bone cement. In the lower left comer of Figure 2(C), small(1-10 pm) BaSO, particles are evident. BaSO, is used to render the bone cement radioopaque. In other areas of the same specimen, the concentration of BaSO, particles was even greater. Areas of BaSO, were seen on the fracture surfaces of at least two other specimens as well. These observations suggest that BaSO, particles adversely affect the fracture resistance of the bone cement in vivo. Several in vitro studies have suggested that a high local concentration of BaSO, particles, due to
146
TOPOLESKI, DUCHEYNE, AND CUCKLER
nonuniform particle distribution, initiates or exacerbates bone cement failure. Bea~mont'~ measured a 40% decrease in fracture toughness, and an increase in fatigue crack velocity in bone cement with the addition of BaSO,. Freitag and CannonI2 measured a decrease of about 10% in bone cement fracture toughness due to the presence of BaSO, particles. Bea~rnont,'~ and Wright and Robinson," observed the formation of voids around the BaSO, particles. Voids surrounding BaSO, particles in vivo are illustrated in Figure 7. The presence of voids surrounding the particles indicates nonbonding of the PMMA to the BaSO,, which means that debonding of the PMMA matrix is not an available energy release mechanism to delay crack propagation. The voids probably act as crack propagation promoters. A feature that was common to the ex uivo crack surfaces was the presence of secondary cracks oriented obliquely to the surface of the primary crack. The stress field within the cement during failure was complex, and multicrack front propagation occurred at several points in the cement mass. Most of the crack fronts (secondary cracks) stabilized, while the energy distribution in the cement mass favored growth of the primary crack, causing cement failure. Figure 8 shows that the secondary cracks were usually associated with pores in the cement mass, and appeared to propagate preferentially
Figure 7. An enlargement of the particles observed on the specimen shown in Figure 2(C) (original magnification lO,OOOX, bar = 1 pm). Voids are clearly surrounding the clumps of BaSOI particles. The formation of the voids around the particles may lead to the lower fracture toughnesses and fatigue lives measured in vitro, and may have contributed to this particular bone cement failure (specimen #11).
PMMA FRACTOGRAPHY
Figure 8. (A) Secondary crack in an ex vim specimen (case #6). The secondary crack is indicated by arrows. The crack has propagated through the pores (original magnification 80X, bar = 100 pm). (B) A secondary crack in another specimen also associated with pores (case #2) (original magnification 160X, bar = 100 pm).
147
148
TOPOLESKI, DUCHEYNE, AND CUCKLER
through the pores. The primary crack differed from the secondary cracks only in that it did not stabilize, but continued to grow. It is likely that the propagation of the primary crack is also influenced by the porosity of the cement. The moving crack was directed toward pores in the cement mass because of the increase of energy available for crack propagation due to the stress concentration effect of the pores. The fractographic observations presented in Figure 8 may provide an explanation of the processes responsible for the increase in fatigue life of PMMA specimens that have been centrifuged to reduce porosity over PMMA specimens that have not been ~ e n t r i f u g e d . ' ~ If~the ~ " stress concentration effect of the pores is eliminated, then a fatigue crack will have less energy available for growth, and therefore growth rate will be reduced. Fractographic phenomena not caused by fatigue were also observed. In one instance (specimen #2) a major element predisposing to rapid fracture was identified. A large pore, shown in Figure 9(A), was clearly associated with the cement failure. The area adjacent to the pore, shown in Figure 9(B), fractured in a cleavage like fashion, similar to the fracture shown in Figure 3(C). The large, elliptical pore significantly weakened the bone cement mantle by reducing the cross section, and generated a stress concentration that eventually initiated an unstable, rapid fracture. The void was probably the cause of the catastrophic failure, based on evidence presented in Figure 9(C), endon views of the void. The edge of the pore is most likely the fracture initiation site, where flat cleavage surfaces, several secondary cracks, and an irregular area at the "cliff-edge" of the pore, are clear. If the void is thought of as an ellipsoidal flaw in the cement mass, the highest stress concentration occurs at the tip of the ellipsoid, the area illustrated in Figure 9(C). Figure 9(D) shows an enlargement of the edge of the void. The noticeable irregularity of the void surface may have been caused by particles of PMMA that did not incorporate well into the bulk of the cement. Poor bonding of the cement particles could have weakened the cement to where the stress concentration effect caused crack initiation. The conclusion remains that the failure of the cement (specimen #2) was due to a rapid fracture, despite uncertainties concerning the cause of fracture initiation. Wear surfaces were observed on several specimens, as illustrated by the two examples of Figure 10. The surfaces are smooth, but differ from the flat, cleavage-like, surfaces in Figures 3(C) and 9(B). The smooth undulations seen in Figure 10 identify the wear surfaces. The surfaces were probably irregular and angular, like the fatigue surfaces presented in Figures 5(C) and 6, and eroded over time. Figure 10(A) was taken from specimen #7, which was known to have a cement failure for at least 3 years prior to revision (illustrated in Fig. 2). Figure 10(B) was taken from specimen #2, known to have a cement failure at least 1 year and 2 months prior to revision. As a fatigue crack progresses, the existing surfaces of the crack will most likely articulate in the cyclic loading environment of a total hip prosthesis, and wear. The presence of wear implies that particles of PMMA have been
PMMA FRACTOGRAPHY
149
(a) Figure 9. (A) Micrograph of the large void in Specimen #2 (original magnification lox, bar = 1 mm). The void spans over 50% of the cement mantle thickness, and was probably the area of initiation of the fracture that lead to the failure of the prosthesis. The area enlarged and presented in (B) is indicated. (8) Smooth cleavage-like fracture indicating a rapid crack growth associated with the pore (original magnification 320X, bar = 50 pm). (C) An end-on view of t h e void shown in (A) (original magnification.40X, bar = 200 pm). A crack is seen moving through the pore surface. Sharp, cleavage-like features are evident just below the edge of the void. An enlargement of the edge of the void, indicated by the arrow (D) shows the details of the possible crack initiation area (original magnification 320X, bar = 50 pm). It appears that there are PMMA particles that were not incorporated into the bulk of the cement, and perhaps weakened the cement and caused crack initiation.
freed, possibly leading to osteolysis and fibrous tissue sheath formation, and enhancing prosthesis loosening. Not all possible combinations of fracture modes have been reproduced in vitro. The complexity of the stress field in viuo obscures the understanding of the exact in vivo fracture modes. One of the principal limitations of the fractographic analysis of ex vivo specimens is that the fracture initiation site rarely can be determined. Slight differences between the ex v i m surface
150
TOPOLESKI, DUCHEYNE, AND CUCKLER
(4 Figure 9. (continued)
PMMA FRACTOGRAPHY
151
(4 Figure 9. (continued)
morphologies and the in vitro surface morphologies, for both the slow and the rapid fractures, are most likely caused by the time the ex vivo specimens remained implanted after cement fracture.: The illumination of the fatigue and slow continuous fracture process of PMMA bone cement explains the origin of the irregular in vivo fracture surface morphology. The similarities between the ex vivo fracture surfaces and the in vitro slow and fatigue fracture surfaces, indicate that the leading cause of bone cement failure in vivo could be either the slow or fatigue fracture. The slow loading rate required to produce a slow continuous failure, however, is not a likely in vivo loading scenario. Bone cement is subjected to cyclic loading in vivo, implicating fatigue crack growth as the most probable leading cause of cement failure. CONCLUSIONS
The present ex vivo and in vitro fractography documents characteristics of in vivo bone cement fracture behavior: 1) Fatigue crack growth is most likely the leading in vivo failure mechanism of bone cement. 2) Both fatigue and slow continuous cracks, in vivo and in vitro, propagate by developing a microcraze shower damage zone.
152
TOPOLESKI, DUCHEYNE, AND CUCKLER
Figure 10. (A) Wear surface found on the specimen #7, illustrated in Figure 1(B) (original magnification 40X, bar = 200 pm). The surface shows the characteristic smoothness and undulations expected of a wear surface. (B) Wear surface found in specimen #2 (original magnification 80X, bar = 100 pm). The undulations are more evident than in (A), which gives the surface the sand dune-like appearance.
PMMA FRACTOGRAPHY
153
3) Agglomerates of BaSO, particles can be implicated in some bone cement failures. 4) Large voids in vivo can lead to a rapid, unstable fracture. 5) Pores in the PMMA mass have a clear influence on a propagating crack. 6) Wear of the fracture surfaces occurs, and may produce PMMA debris, exacerbating bone destruction. The data presented suggest that methods to prevent subcritical crack propagation are important for improving the mechanical performance of PMMA bone cement. The Authors would like to thank: XiCi Lu, a visiting scientist from the People‘s Republic of China, for his valuable discussions regarding polymer fracture; R. Heppenstall, M.D., J. Esterhai, M.D., and M. Steinberg, M.D., for providing some of the in vivo fractured cement specimens; and Howmedica, Inc., for a donation of bone cement. This research was supported in part by the State of Pennsylvania through a Benjamin Franklin Partnership Grant.
References 1. M. B. Coventry, R. D. Beckenbaugh, D. R. Nolan, and D. M. Ilstrup, 2.
3. 4.
5. 6.
7. 8. 9.
10. 11. 12.
“2012 total hip arthroplasties: A study of postoperative course and early complications,” J. Bone J. Surg., 56-A, 273-284 (1974). P. M. Pellicci, P. D. Wilson, Jr., C. B. Sledge, E. A. Salvati, C. S. Ranawat, and R. Poss, ”Revision total hip arthroplasty,” Clin. Orthop., 170, 34-41 (1982). E. T. R. James, G. A. Hunter, and H. U. Cameron, ”Total hip revision arthroplasty: Does sepsis influence the results,” Clin. Orthop., 170,88-94 (1982). H. I?. Chandler, F.T. Reineck, R. L. Wixon, and J. C. McCarthy, “Total hip replacement in patients younger than thirty years old. A five year follow-up study,” J. Bone J. Surg., 63-A, 1426-1434 (1981). L. D. Dorr, G . K. Takei, and J. P. Conaty, “Total hip arthroplasties in patients less than forty-five years old,” J. Bone J. Surg., 65-A, 474-479 (1983). S. H. White, ”The fate of cemented total hip arthroplasty in young patients,” Clin. Orthop., 231, 29-34 (1988). R. N. Stauffer, “Ten-year follow-up study of total hip replacement,” 1. Bone 1. Surg., @-A, 983-990 (1982). T. A. Gmen, G. M. NcNeice, and H. C. Amstutz, “‘Modes of failure’ of cemented stem-type femoral components,” Clin. Orthop., 141, 17-27 (1979). H. J. Agins, N. W. Alcock, B. Manjula, E. A . Salvati, P. D. Wilson, P. M. Pellicci, and P. G. Bullough, ”Metallic wear in failed titaniumalloy total hip replacements,” J. Bone J. Surg., 70-A, 347-356 (1988). H. G. Willert, K. Mueller, and M. Semlitsch, ”The morphology of polymethylmethacrylate (PMMA) bone cement surface structures and causes of their origin,” Arch. Ortho. Traumat. Surg., 94, 265-292 (1979). T. A. Freitag, and S. L. Cannon, “Fracture characteristics of acrylic bone cements 11. Fatigue,” J. Biomed. Muter. Res., 11, 609-624 (1977). T. A. Freitag, and S. L. Cannon, “Fracture characteristics of acrylic bone cements I. Fracture toughness,” 1. Biomed. Muter. Res., 10, 805-828 (1977).
TOPOLESKI, DUCHEYNE, AND CUCKLER
154
13. P. W. R. Beaumont, “The strength of acrylic bone cements and acrylic
14. 15. 16.
17. 18.
19. 20.
21. 22.
23. 24.
25. 26. 27. 28. 29.
30.
cement-stainless steel interfaces. Part 1. The strength of acrylic cement containing second phase dispersions,“ J. Muter. Sci., 12, 1845-1852 (1977). D. W. Burke, E. I. Gates, and W. H. Harris, ”Centrifugation as a method of improving tensile and fatigue properties of acrylic bone cements,” J. Bone J. Surg., 66-A, 1265-1273 (1984). R. M. Pilliar, R. Blackwell, I. McNab, and H. U. Cameron, ”Carbon fiber-reinforced bone cement in orthopedic surgery,” J. Biomed. Mater. Res., 10, 893-906 (1976). R. M. Pilliar, W. J. Bratina, and R. A. Blackwell, ”Mechanical properties of carbon fiber-reinforced polymethylmethacrylate for surgical implant applications,“ in Fatigue of Filamentay Composite Materials, ASTM STP 636, K. L. Reifsnider, and K. N. Lauraitis (eds.), ASTM, Philadelphia, 1977, pp. 206-227. C. M. Rimnac, T. M. Wright, and D. L. McGill, “The effect of centrifugation on the fracture properties of acrylic bone cements,” J. Bone J. Surg., 68-A, 281-287 (1986). T. M. Wright and R. P. Robinson, “Fatigue crack propagation in polymethylmethacrylate bone cements,” f. Muter. Sci., 17, 2463-2468 (1982). F. A. Weber and J. Charnley, “A radiological study of fractures of acrylic cement in relation to the stem of a femoral head prosthesis,” J. Bone J. Surg., 59-€3, 297-301 (1975). R. Schneider, R. H. Freiberger, B. Ghelman, and C. S. Ranawat, ”Radiological evaluation of painful joint prostheses,” Clin. Orthop., 170, 156168 (1982). P. W. R. Beaumont and R. J. Young, ”Slow crack growth in acrylic bone cement,” J. Biomed. Mater. Res., 9, 423-439 (1975). R. P. Kusy and D. T. Turner, “Influence of the molecular weight of poly (methyl methacrylate) on fracture morphology in notched tension,” Polymer, 18, 391-402 (1977). X. Lu, ”The fracture toughness and fractographies of acrylics with different compositions,” Proceedings of the Binary Polymer Science Conference of China and Japan, Beijing, China, Nov. 1983, pp. 405-406. W. Doll, ”Optical interference measurements and fracture mechanics analysis of crack tip craze zones,” in Advances in Polymer Sciences, 52153, Crazing in Polymers, H. H. Kausch, (ed.) Springer-Verlag, New York, 1983, pp. 105-168. I. Wolock and S. B. Newman, “Fracture topography,” in Fracture Processes in Polymeric Solids, Phenomena and Theory, B. Rosen, (ed.), Interscience, New York, 1964, pp. 235-290. J. P. Berry, ”Genera1 theory of brittle fracture,” in Fracture Processes in Polymeric Solids, Phenomena and Theory, B. Rosen (ed.), Interscience, New York, 1964, pp. 157-234. R. P. Kusy and D. T. Turner, “Fractography of poly (methyl methacrylates),” J. Biomed. &later. Res. Symp., 6, 89-98 (1975). W. Doll, “A molecular weight dependent fracture transition in poly methylmethacrylate,” 1. Mater. Sci., 10, 935-942 (1975). S. L. Kim, M. Skibo, J. A. Manson, and R. W. Hertzberg, “Fatigue crack propagation in poly (methyl methacrylate): Effect of molecular weight and internal plasticization,” Potym. Eng. Sci., 17, 194-203 (1977). J. P. Davies, D. 0. OConnor, D. W. Burke, M. Jasty, and W. H. Hams, “The effect of centrifugation on the fatigue life of bone cement in the presence of surface irregularities,” Clin. Orthop., 229, 156-161 (1988).
Received December, 1988 Accepted July 17, 1989
elucidation of PMMA fracture micromechanisms. The ex viuo fracture surfaces are remarkably similar to in vitro fatigue fracture surfaces. The fractographic data document that the primary in vivo failure mechanism of bone cement is fatigue, and the fatigue cracks grow by developing a microcraze shower damage zone. Agglomerates of BaS04 particles can be implicated in some bone cement failures, large flaws or voids in vivo can lead to a rapid, unstable fracture, pores in the PMMA mass have a clear influence on a propagating crack, and wear of the fracture surfaces occurs, and may produce PMMA debris, exacerbating bone destruction.
INTRODUCTION
Cementing with poly(methy1 methacrylate) (PMMA) is the predominant means of fixing total hip prostheses. The success rate of primary cemented hip arthroplasties at ten years exceeds 90% in patients 60 or more years of age when proper cementing techniques are however, bone cement does fail mechanically. In addition, the incidence of cemented prosthesis failure in patients younger than fifty is considerably higher.44 Failure is often associated with several clinically observed modes of prosthesis loosening.7,8The mechanical failure of bone cement can result further in the production of cement particles. The particles interact with the surrounding Journal of Biomedical Materials Research, Vol. 24, 135-154 (1990) CCC 0021-93O4/90/.020135-20$04.00 0 1990 John Wiley & Sons, Inc.
136
TOPOLESKI, DUCHEYNE, AND CUCKLER
tissues, and may contribute to an inflammatory response, increasing bone destruction, and accelerating prosthesis loosening. Prosthesis loosening frequently requires correction via revision surgery. The improvement of the bone cement integrity is fundamental to the longevity of cemented prosthesis systems, The identification of properties critical to the in vivo performance of bone cement will focus bone cement optimization efforts. Failed bone cement specimens retrieved from patients provide the unique source of evidence for studying in-service failure mechanisms of this material. Fractographic analysis of retrieved failed cement is the most useful method for observing failure patterns and establishing significant mechanical properties. Few studies have reported analyses of bone cement specimens retrieved from in vivo service, Willert et al.'" presented scanning electron microscopy (SEM) data describing the morphology of bone cement retrieved from humans. They examined the endosteal surfaces of the PMMA, and did not consider in vivo fracture surfaces of bone cement. Bone cement fractography has been reported infrequently, primarily as an adjunct to in vitro bone cement material characterization studies. Freitag and Cannon'' identified the extent of fatigue crack growth before catastrophic failure by observing a rough to smooth transition of the fracture surface. A review of other bone cement fractography12-18shows that distinct fracture surface morphologies are associated with the several modes of fracture possible in materials. The mechanisms of bone cement failure in vivo have not been investigated in depth; thus, our objective was to elucidate the in vivo failure mechanisms of bone cement. Fracture surfaces of bone cement specimens that failed in vivo, were compared with fatigue, slow continuous, and rapid continuous fracture surfaces created in vitro. In vivo fracture surface morphologies of bone cement were identified and related to PMMA fracture micromechanisms. METHODS AND MATERIALS
Specimens of PMMA, from 12 failed cemented arthroplasties, were obtained during revision surgery. The cement fractures were generally visible in preoperative radiographs, except in three cases. Clinical data corresponding to each specimen of fractured cement are summarized in Table I. The average age of the patients included in this study was 73 years. The average time between initial implant and revision was 8 years, 9 months. Figure 1(A) illustrates a radiograph showing multiple cement fractures. Loosening and subsequent revision of the illustrated prosthesis (case #7) was performed 3 years after cement fractures were first observed radiographically. Six of the prostheses that were revised and contributed bone cement to this study showed only single, longitudinal bone cement fractures in preoperative radiographs. In contrast to an observation often made in the l i t e r a t ~ r e , 'these ~ , ~ ~ cases were revised for loosening.
PMMA FRACTOGRAPHY
137
TABLE I Clinical Data: In V i m Specimen Sources
Specimen
Age at Revision 63 81 81 71 61
6 7 8 9
84
10
45
11 12
62 69
80 66 73
Sex F F F M M F F F F
M M M
Time between Initial Surgery and Revision 13 years, 6 years, 2 years, 14 years, 2 years,
Time between Cement Failure Diagnosis and Revision
9 months
a.
5 months 11 months 7 months 7 months
1year, 2 months 1 year, 5 months
unknown 7 years, 0 months 14 years, 0 months 2 years, 2 months 14 years, 0 months 2 years, 9 months 6 years, 5 months
0 0
unknown 3 years, 0 months 1 yearb 2 months unknown 2 years 3 years
Initial Implant Type Charnley Charnley Mueller Mueller Bipolar (long stem) T 28 Mueller Mueller Mueller Bipolar Mueller Mueller
aZero (0) time indicates cement failure diagnosed at surgery. bAbsence of specific number of months indicates record document only the year involved.
Only pieces of fractured cement that were removed easily from the medullary canal were included in the study; thus artifacts on the fracture surfaces caused by instrument damage were avoided. All specimens were cleaned in a detergent* to remove biological debris. They were then cut using a slow-speed diamond saw, and sputter-coated with gold. A photograph of a prepared fracture specimen (from case #7) is shown in Figure l(B). A commercially available bone cementt was used to make the in vitro specimens. The monomer was chilled at O°C at least 20 min prior to mixing. The bone cement was mixed according to the instructions of the package insert. Fracture surfaces were created irt vitvo by four distinct methods: 1) Fatigue surfaces, Type I, were produced by fatiguing rectangular prismatic bars (10 X 10 x 60 mm) in three-point bending. A 2.0-mm-long notch was machined at the specimen mid-span with a low-speed diamond saw, and then a sharp groove was cut into the tip of the notch with a scalpel. A fatigue crack approximately 2.5-3.5 mm long was propagated at 10 Hz from the notch tip. Fatiguing was performed under stroke controls at a load that cycled between 22 and 133 N (5-30 lbs.) 2) and 3) Rotating-bending fatigue surfaces, both high (Type 11) and low (Type 111) cycle: Cylindrical waisted beams (5-mm waist diameter) were subjected to fully reversed rotating-bending at a frequency of 30 H Z .The ~ maxi*Liqui-Nox: Alconox, Inc., New York, NY. 'Simplex P: Howmedica, Inc., Rutherford, N.J. $Instron 1331: Instron, Inc., Canton, MA. $RBF-200: Fatigue Dynamics, Inc., Dearborn, MI.
138
TOPOLESKI, DUCHEYNE, AND CUCKLER
(a) Figure 1. (A) Preoperative (revision) radiograph of case #7 (Table I). (B) Photograph of one of the specimens of case #7 prepared for analysis in the SEM. The specimen origin is indicated by the arrow in (A).
mum stresses applied were 9.0 and 17.2 MPa, respectively. The specimens were stored in physiological saline at 37°C for at least 1 week prior to machining, and then replaced into saline at 37°C for at least 1 additional week. 4) Slow continuous fracture surfaces were produced by loading the Type I fatigue specimens at a constant displacement rate of 0.038 mm/s, without fatiguing. The effect of loading the specimen at a slow loading rate, with only the scalpel cut as a starter notch, resulted in initial slow crack growth, followed by an unstable, rapid crack growth. 5) Rapid continuous fracture surfaces were created by loading the Type I specimens at constant displacement rate of 0.076 mm/s, until unstable failure occurred. In view of the studies of Freitag and Cannon11'12 and Beaumont,"13 who documented the effect of test environment on static and fatigue testing in
139
PMMA FRACTOGRAPHY
VJ) Figure 1. (continued)
PMMA bone cement, static tests were performed in air. Fatigue test specimens were maintained constantly moist with physiological saline. RESULTS AND DISCUSSION
The surfaces observed in the majority of the ex vivo analyses were irregular and rough. Example of ex vim surfaces from three separate specimens are included in Figure 2. The morphology of the surfaces resembles irregular stacks of plates or plateaus, and is not peaked or mountainous. Features typical of each of the in vitvo fracture surfaces are shown in Figures 3 and 4. Figure 3(A) shows the overall fracture surface typical of the fracture toughness specimens. The fatigue surface appears as a light colored
140
TOPOLESKI, DUCHEYNE, AND CUCKLER
(4 Figure 2. Ex uiuo fracture surfaces from three different specimens. (A) specimen #6, (B) specimen #7, and (C) specimen #11. Each of the surfaces is irregular and stepped (original magnification 320X, bar = 50 ,urn).
141
PMMA FRACTOGRAPHY
NOTCH
+ t w
FRACTURE
(a)
Figure 3. (A) Overall view of the fracture surface from the fracture toughness preparation. The fatigue surface appears as a light-colored parabolic, or "thumbnail," shaped surface contrasting against the darker, flat, rapid fracture surface. Arrow indicates direction of crack propagation (original magnification 1OX). (B) Enlargement of the fatigue surface. The surface is similar to the ex v i m surfaces shown in Figure 3 . (original magnification 320X, bar = 50 pm). (C) An enlargement of the rapid fracture surface. The twophase nature of bone cement is evident by the clear definition of the beads that make up the powder component of PMMA (original magnification 320X, bar = 50 pm).
parabolic, or "thumbnail", shaped surface contrasting against the darker, flat, rapid fracture surface. Figure 3(B) shows an enlargement of the Type I fatigue surface. Figure 3(C) shows an enlargement of the smooth fracture surface typical of the rapid, unstable, crack growth. The two-phase nature of bone cement is evident by the clear definition of the beads that make up the powder component of PMMA. Figure 4 shows the irregular surface typical of both Type I1 and Type I11 fatigue fracture surfaces. The three types of in vitvo fatigue surfaces are similar. The slow continuous fracture surface was virtually identical to the fatigue surfaces. The similarity of the Type I, 11, and I11 fatigue, and the slow continuous fracture surfaces indicate that the microstructural mechanism causing failure was the same in each case.
142
TOPOLESKI, DUCHEYNE, AND CUCKLER
(c)
Figure 3. (continued)
The ex vivo fracture surfaces are remarkably similar to the in vitro fatigue fracture surfaces. Each of the ex vivo surfaces illustrated in Figure 2 exhibits
PMMA FRACTOGRAPHY
143
Figure 4. Micrograph of the irregular surface associated with Type I1 fatigue cracks (original magnification 320X, bar = 50 Pm). The Type I1 and Type I11 fatigue surfaces are indistinguishable from the Type I fatigue surface (Fig. 3(B)) and the ex uiuo fracture surfaces (Fig. 2).
the stepped surface irregularities that are clear in Figures 3(B) and 4. The proposed crack extension process responsible for the irregular fatigue surface morphology found both ex vivo and in vitra is illustrated by the micrograph presented in Figure 5. The micrograph shows a side view of a three-point bending specimen, looking at the specimen perpendicular to the plane of crack growth. The process illustrated is known as a “microcraze shower,” and occurs when the molecular weight of the PMMA is too low to form fibrils, or craze, across the crack tip damage zone (Fig. 6(A)).22,23 The microcraze shower forms the damage zone ahead of the crack tip (Fig. 6(B)). As energy is added to the system, the microcrazes are connected, and the crack extends (Fig. 6(C)). The characteristic irregular fatigue surface morphology may be the product of the microcraze coalescence. The smooth, rapid fracture surface shown in Figure 3(C) indicates that no microcraze damage zone was formed. The two-phase nature of PMMA bone cement is clear in Figure 3(C). Clawlike markings are observed, but are confined to the fracture path through the polymer beads. The clawlike markings are typical of polymer fracture surfaces where the polymer is of sufficiently high molecular weight, and indicates a layer of crazed The moIecular weight of each phase of the bone cement governs the fracture mechanism, and hence the fracture morphology of the specimens. The material rupture within the mi-
144
TOPOLESKI, DUCHEYNE, AND CUCKLER
Figure 5. Micrograph of the Type I fatigue crack tip. The crack propagation direction is indicated by the black arrow. The microcrazes ahead of the crack tip a r e indicated by t h e white arrow (original magnification 8OX, bar = 100 pm).
crocraze shower probably occurs preferentially through the binding phase PMMA. In addition to the low fatigue crack propagation resistance of the binding phase, the fracture surfaces also evidenced that the prepolymerized phase (the beads) display a relatively low crack propagation resistance, typical for a low molecular weight PMMA.24-26,2s,29 High molecular weight poly(methy1 methacrylate)s, such as industrial PMMA, exhibit fatigue striations. The striations are associated with specific fatigue crack propagation mechanisms. Kim et a1.28showed that fatigue crack propagation continues to decrease with increasing molecular weight even at molecular weights exceeding the maximum molecular weight for which there is a fracture toughness molecular weight dependence. This observation, coupled with the findings of Beaumont and Young,*l suggests that fatigue fracture surface topographies are dependent on crack propagation mechanisms, which, in turn, depend on molecular weight. Beaumont and Youn$ showed that, indeed, the slow fracture surface of high-molecular weight PMMA (viscosity average molecu-
145
PMMA FRACTOGRAPm
A:
6:
+CRACK
__ct)_
DAMAbE
ZONE
(mitrot rare)
C:
Figure 6. A schematic of possible crack propagation mechanisms in PMMA bone cement. (A) If the polymer is of sufficiently high molecular weight, a damage zone will form where fibrils span the crack surfaces. When the stress across the fibrils is great enough, the fibrils will break, and the crack extends. (B) If the molecular weight of the polymer is not high enough to form fibrils across the damage zone, the crack tip may be preceded by a microcraze shower. (C) As the crack opens, the microcrazes coalesce and the crack extends.
lar weight ( M J > 1,000,000) had a smooth appearance, while the slow fracture surface of bone cement (M, = 170,000), was irregular. Neither the ex vivo nor the in vitm fatigue surfaces observed in this study exhibited fatigue striations associated with high molecular weight PMMA. There were several other important features identified through the ex vivo fractography that pertain to the in vivo failure of bone cement. In the lower left comer of Figure 2(C), small(1-10 pm) BaSO, particles are evident. BaSO, is used to render the bone cement radioopaque. In other areas of the same specimen, the concentration of BaSO, particles was even greater. Areas of BaSO, were seen on the fracture surfaces of at least two other specimens as well. These observations suggest that BaSO, particles adversely affect the fracture resistance of the bone cement in vivo. Several in vitro studies have suggested that a high local concentration of BaSO, particles, due to
146
TOPOLESKI, DUCHEYNE, AND CUCKLER
nonuniform particle distribution, initiates or exacerbates bone cement failure. Bea~mont'~ measured a 40% decrease in fracture toughness, and an increase in fatigue crack velocity in bone cement with the addition of BaSO,. Freitag and CannonI2 measured a decrease of about 10% in bone cement fracture toughness due to the presence of BaSO, particles. Bea~rnont,'~ and Wright and Robinson," observed the formation of voids around the BaSO, particles. Voids surrounding BaSO, particles in vivo are illustrated in Figure 7. The presence of voids surrounding the particles indicates nonbonding of the PMMA to the BaSO,, which means that debonding of the PMMA matrix is not an available energy release mechanism to delay crack propagation. The voids probably act as crack propagation promoters. A feature that was common to the ex uivo crack surfaces was the presence of secondary cracks oriented obliquely to the surface of the primary crack. The stress field within the cement during failure was complex, and multicrack front propagation occurred at several points in the cement mass. Most of the crack fronts (secondary cracks) stabilized, while the energy distribution in the cement mass favored growth of the primary crack, causing cement failure. Figure 8 shows that the secondary cracks were usually associated with pores in the cement mass, and appeared to propagate preferentially
Figure 7. An enlargement of the particles observed on the specimen shown in Figure 2(C) (original magnification lO,OOOX, bar = 1 pm). Voids are clearly surrounding the clumps of BaSOI particles. The formation of the voids around the particles may lead to the lower fracture toughnesses and fatigue lives measured in vitro, and may have contributed to this particular bone cement failure (specimen #11).
PMMA FRACTOGRAPHY
Figure 8. (A) Secondary crack in an ex vim specimen (case #6). The secondary crack is indicated by arrows. The crack has propagated through the pores (original magnification 80X, bar = 100 pm). (B) A secondary crack in another specimen also associated with pores (case #2) (original magnification 160X, bar = 100 pm).
147
148
TOPOLESKI, DUCHEYNE, AND CUCKLER
through the pores. The primary crack differed from the secondary cracks only in that it did not stabilize, but continued to grow. It is likely that the propagation of the primary crack is also influenced by the porosity of the cement. The moving crack was directed toward pores in the cement mass because of the increase of energy available for crack propagation due to the stress concentration effect of the pores. The fractographic observations presented in Figure 8 may provide an explanation of the processes responsible for the increase in fatigue life of PMMA specimens that have been centrifuged to reduce porosity over PMMA specimens that have not been ~ e n t r i f u g e d . ' ~ If~the ~ " stress concentration effect of the pores is eliminated, then a fatigue crack will have less energy available for growth, and therefore growth rate will be reduced. Fractographic phenomena not caused by fatigue were also observed. In one instance (specimen #2) a major element predisposing to rapid fracture was identified. A large pore, shown in Figure 9(A), was clearly associated with the cement failure. The area adjacent to the pore, shown in Figure 9(B), fractured in a cleavage like fashion, similar to the fracture shown in Figure 3(C). The large, elliptical pore significantly weakened the bone cement mantle by reducing the cross section, and generated a stress concentration that eventually initiated an unstable, rapid fracture. The void was probably the cause of the catastrophic failure, based on evidence presented in Figure 9(C), endon views of the void. The edge of the pore is most likely the fracture initiation site, where flat cleavage surfaces, several secondary cracks, and an irregular area at the "cliff-edge" of the pore, are clear. If the void is thought of as an ellipsoidal flaw in the cement mass, the highest stress concentration occurs at the tip of the ellipsoid, the area illustrated in Figure 9(C). Figure 9(D) shows an enlargement of the edge of the void. The noticeable irregularity of the void surface may have been caused by particles of PMMA that did not incorporate well into the bulk of the cement. Poor bonding of the cement particles could have weakened the cement to where the stress concentration effect caused crack initiation. The conclusion remains that the failure of the cement (specimen #2) was due to a rapid fracture, despite uncertainties concerning the cause of fracture initiation. Wear surfaces were observed on several specimens, as illustrated by the two examples of Figure 10. The surfaces are smooth, but differ from the flat, cleavage-like, surfaces in Figures 3(C) and 9(B). The smooth undulations seen in Figure 10 identify the wear surfaces. The surfaces were probably irregular and angular, like the fatigue surfaces presented in Figures 5(C) and 6, and eroded over time. Figure 10(A) was taken from specimen #7, which was known to have a cement failure for at least 3 years prior to revision (illustrated in Fig. 2). Figure 10(B) was taken from specimen #2, known to have a cement failure at least 1 year and 2 months prior to revision. As a fatigue crack progresses, the existing surfaces of the crack will most likely articulate in the cyclic loading environment of a total hip prosthesis, and wear. The presence of wear implies that particles of PMMA have been
PMMA FRACTOGRAPHY
149
(a) Figure 9. (A) Micrograph of the large void in Specimen #2 (original magnification lox, bar = 1 mm). The void spans over 50% of the cement mantle thickness, and was probably the area of initiation of the fracture that lead to the failure of the prosthesis. The area enlarged and presented in (B) is indicated. (8) Smooth cleavage-like fracture indicating a rapid crack growth associated with the pore (original magnification 320X, bar = 50 pm). (C) An end-on view of t h e void shown in (A) (original magnification.40X, bar = 200 pm). A crack is seen moving through the pore surface. Sharp, cleavage-like features are evident just below the edge of the void. An enlargement of the edge of the void, indicated by the arrow (D) shows the details of the possible crack initiation area (original magnification 320X, bar = 50 pm). It appears that there are PMMA particles that were not incorporated into the bulk of the cement, and perhaps weakened the cement and caused crack initiation.
freed, possibly leading to osteolysis and fibrous tissue sheath formation, and enhancing prosthesis loosening. Not all possible combinations of fracture modes have been reproduced in vitro. The complexity of the stress field in viuo obscures the understanding of the exact in vivo fracture modes. One of the principal limitations of the fractographic analysis of ex vivo specimens is that the fracture initiation site rarely can be determined. Slight differences between the ex v i m surface
150
TOPOLESKI, DUCHEYNE, AND CUCKLER
(4 Figure 9. (continued)
PMMA FRACTOGRAPHY
151
(4 Figure 9. (continued)
morphologies and the in vitro surface morphologies, for both the slow and the rapid fractures, are most likely caused by the time the ex vivo specimens remained implanted after cement fracture.: The illumination of the fatigue and slow continuous fracture process of PMMA bone cement explains the origin of the irregular in vivo fracture surface morphology. The similarities between the ex vivo fracture surfaces and the in vitro slow and fatigue fracture surfaces, indicate that the leading cause of bone cement failure in vivo could be either the slow or fatigue fracture. The slow loading rate required to produce a slow continuous failure, however, is not a likely in vivo loading scenario. Bone cement is subjected to cyclic loading in vivo, implicating fatigue crack growth as the most probable leading cause of cement failure. CONCLUSIONS
The present ex vivo and in vitro fractography documents characteristics of in vivo bone cement fracture behavior: 1) Fatigue crack growth is most likely the leading in vivo failure mechanism of bone cement. 2) Both fatigue and slow continuous cracks, in vivo and in vitro, propagate by developing a microcraze shower damage zone.
152
TOPOLESKI, DUCHEYNE, AND CUCKLER
Figure 10. (A) Wear surface found on the specimen #7, illustrated in Figure 1(B) (original magnification 40X, bar = 200 pm). The surface shows the characteristic smoothness and undulations expected of a wear surface. (B) Wear surface found in specimen #2 (original magnification 80X, bar = 100 pm). The undulations are more evident than in (A), which gives the surface the sand dune-like appearance.
PMMA FRACTOGRAPHY
153
3) Agglomerates of BaSO, particles can be implicated in some bone cement failures. 4) Large voids in vivo can lead to a rapid, unstable fracture. 5) Pores in the PMMA mass have a clear influence on a propagating crack. 6) Wear of the fracture surfaces occurs, and may produce PMMA debris, exacerbating bone destruction. The data presented suggest that methods to prevent subcritical crack propagation are important for improving the mechanical performance of PMMA bone cement. The Authors would like to thank: XiCi Lu, a visiting scientist from the People‘s Republic of China, for his valuable discussions regarding polymer fracture; R. Heppenstall, M.D., J. Esterhai, M.D., and M. Steinberg, M.D., for providing some of the in vivo fractured cement specimens; and Howmedica, Inc., for a donation of bone cement. This research was supported in part by the State of Pennsylvania through a Benjamin Franklin Partnership Grant.
References 1. M. B. Coventry, R. D. Beckenbaugh, D. R. Nolan, and D. M. Ilstrup, 2.
3. 4.
5. 6.
7. 8. 9.
10. 11. 12.
“2012 total hip arthroplasties: A study of postoperative course and early complications,” J. Bone J. Surg., 56-A, 273-284 (1974). P. M. Pellicci, P. D. Wilson, Jr., C. B. Sledge, E. A. Salvati, C. S. Ranawat, and R. Poss, ”Revision total hip arthroplasty,” Clin. Orthop., 170, 34-41 (1982). E. T. R. James, G. A. Hunter, and H. U. Cameron, ”Total hip revision arthroplasty: Does sepsis influence the results,” Clin. Orthop., 170,88-94 (1982). H. I?. Chandler, F.T. Reineck, R. L. Wixon, and J. C. McCarthy, “Total hip replacement in patients younger than thirty years old. A five year follow-up study,” J. Bone J. Surg., 63-A, 1426-1434 (1981). L. D. Dorr, G . K. Takei, and J. P. Conaty, “Total hip arthroplasties in patients less than forty-five years old,” J. Bone J. Surg., 65-A, 474-479 (1983). S. H. White, ”The fate of cemented total hip arthroplasty in young patients,” Clin. Orthop., 231, 29-34 (1988). R. N. Stauffer, “Ten-year follow-up study of total hip replacement,” 1. Bone 1. Surg., @-A, 983-990 (1982). T. A. Gmen, G. M. NcNeice, and H. C. Amstutz, “‘Modes of failure’ of cemented stem-type femoral components,” Clin. Orthop., 141, 17-27 (1979). H. J. Agins, N. W. Alcock, B. Manjula, E. A . Salvati, P. D. Wilson, P. M. Pellicci, and P. G. Bullough, ”Metallic wear in failed titaniumalloy total hip replacements,” J. Bone J. Surg., 70-A, 347-356 (1988). H. G. Willert, K. Mueller, and M. Semlitsch, ”The morphology of polymethylmethacrylate (PMMA) bone cement surface structures and causes of their origin,” Arch. Ortho. Traumat. Surg., 94, 265-292 (1979). T. A. Freitag, and S. L. Cannon, “Fracture characteristics of acrylic bone cements 11. Fatigue,” J. Biomed. Muter. Res., 11, 609-624 (1977). T. A. Freitag, and S. L. Cannon, “Fracture characteristics of acrylic bone cements I. Fracture toughness,” 1. Biomed. Muter. Res., 10, 805-828 (1977).
TOPOLESKI, DUCHEYNE, AND CUCKLER
154
13. P. W. R. Beaumont, “The strength of acrylic bone cements and acrylic
14. 15. 16.
17. 18.
19. 20.
21. 22.
23. 24.
25. 26. 27. 28. 29.
30.
cement-stainless steel interfaces. Part 1. The strength of acrylic cement containing second phase dispersions,“ J. Muter. Sci., 12, 1845-1852 (1977). D. W. Burke, E. I. Gates, and W. H. Harris, ”Centrifugation as a method of improving tensile and fatigue properties of acrylic bone cements,” J. Bone J. Surg., 66-A, 1265-1273 (1984). R. M. Pilliar, R. Blackwell, I. McNab, and H. U. Cameron, ”Carbon fiber-reinforced bone cement in orthopedic surgery,” J. Biomed. Mater. Res., 10, 893-906 (1976). R. M. Pilliar, W. J. Bratina, and R. A. Blackwell, ”Mechanical properties of carbon fiber-reinforced polymethylmethacrylate for surgical implant applications,“ in Fatigue of Filamentay Composite Materials, ASTM STP 636, K. L. Reifsnider, and K. N. Lauraitis (eds.), ASTM, Philadelphia, 1977, pp. 206-227. C. M. Rimnac, T. M. Wright, and D. L. McGill, “The effect of centrifugation on the fracture properties of acrylic bone cements,” J. Bone J. Surg., 68-A, 281-287 (1986). T. M. Wright and R. P. Robinson, “Fatigue crack propagation in polymethylmethacrylate bone cements,” f. Muter. Sci., 17, 2463-2468 (1982). F. A. Weber and J. Charnley, “A radiological study of fractures of acrylic cement in relation to the stem of a femoral head prosthesis,” J. Bone J. Surg., 59-€3, 297-301 (1975). R. Schneider, R. H. Freiberger, B. Ghelman, and C. S. Ranawat, ”Radiological evaluation of painful joint prostheses,” Clin. Orthop., 170, 156168 (1982). P. W. R. Beaumont and R. J. Young, ”Slow crack growth in acrylic bone cement,” J. Biomed. Mater. Res., 9, 423-439 (1975). R. P. Kusy and D. T. Turner, “Influence of the molecular weight of poly (methyl methacrylate) on fracture morphology in notched tension,” Polymer, 18, 391-402 (1977). X. Lu, ”The fracture toughness and fractographies of acrylics with different compositions,” Proceedings of the Binary Polymer Science Conference of China and Japan, Beijing, China, Nov. 1983, pp. 405-406. W. Doll, ”Optical interference measurements and fracture mechanics analysis of crack tip craze zones,” in Advances in Polymer Sciences, 52153, Crazing in Polymers, H. H. Kausch, (ed.) Springer-Verlag, New York, 1983, pp. 105-168. I. Wolock and S. B. Newman, “Fracture topography,” in Fracture Processes in Polymeric Solids, Phenomena and Theory, B. Rosen, (ed.), Interscience, New York, 1964, pp. 235-290. J. P. Berry, ”Genera1 theory of brittle fracture,” in Fracture Processes in Polymeric Solids, Phenomena and Theory, B. Rosen (ed.), Interscience, New York, 1964, pp. 157-234. R. P. Kusy and D. T. Turner, “Fractography of poly (methyl methacrylates),” J. Biomed. &later. Res. Symp., 6, 89-98 (1975). W. Doll, “A molecular weight dependent fracture transition in poly methylmethacrylate,” 1. Mater. Sci., 10, 935-942 (1975). S. L. Kim, M. Skibo, J. A. Manson, and R. W. Hertzberg, “Fatigue crack propagation in poly (methyl methacrylate): Effect of molecular weight and internal plasticization,” Potym. Eng. Sci., 17, 194-203 (1977). J. P. Davies, D. 0. OConnor, D. W. Burke, M. Jasty, and W. H. Hams, “The effect of centrifugation on the fatigue life of bone cement in the presence of surface irregularities,” Clin. Orthop., 229, 156-161 (1988).
Received December, 1988 Accepted July 17, 1989
