Polycrystalline Zirconia For Joint Replacement

Hund first drew the image of a completely filled cationic but deficient anionic sublattice in 1951 [574]. Nowadays, the possibility of alloying ZrO2 by doping with Y3+ ions represents an essential discovery in bioceramic science since ZrO2 (partly) stabilized by Y2O3 is one of the most commonly used ceramic oxides in joint arthroplasty. Through addition of Y2O3 to the ZrO2 structure, oxygen vacancies can form to achieve charge neutrality in the crystal according to the following stoichiometric reaction:x2 3 Zr O OY O 2Y 3O V••→ + +′ (4.1)where, x2 3 Zr O OY O 2Y 3O V••→ + +′ represents the Y ions on regular cationic sites (i.e., replacing Zr4+ ions) with a relative charge of –1; x2 3 Zr O OY O 2Y 3O V••→ + +′ stands for oxygen ions on regular anionic sites and x2 3 Zr O OY O 2Y 3O V••→ + +′ represents the double positively charged (with respect to the charge of regular oxygen ions) oxygen vacancies. The Kröger-Vink notation for lattice elements and point defects in crystal structures is used here (and will be used throughout the all book) [575], in order to describe the atomistic reactions related to the replacement of Zr4+ by Y3+. Note that two dopant Y ions are necessary to introduce a single oxygen vacancy. Accordingly, the concentration of vacancies on the anionic sublattice can be calculated from the concentration of Y ions on the cationic one and might easily reach the range of few percent, as follows:O Zr1V Y2••   = ′    (4.2) The concentration of intrinsic defects thermally induced during sintering is, in contrast, negligible in the usual range of manufacturing temperatures, which explains the full instability of pure zirconia in its tetragonal phase at room temperature. Note that the amount of doping needed for full stabilization is quite substantial (>8 mol.% for various dopants in order to achieve full stabilization [576-578]), thus showing that the electrostatic neutrality of this rather ionic material can only be maintained by large amounts of oxygen vacancies. The simultaneous presence of dopant cations and oxygen vacancies in large concentration also means that the local atomic environments in the stabilized material are very different from the corresponding stoichiometric phases. Unfortunately, there is yet no clear picture of the mechanisms of stabilization on the atomic scale, the most relevant issues concerning

the roles of different impurity cations and of oxygen vacancies [579]. In particular, the position of the oxygen vacancy associated with each pair of Y atoms has been the subject of deep discussions. Preliminary experimental data appeared to support the hypothesis that Y is nearest neighbor to the oxygen vacancy (Fig. 4.1(d)) [580-583]; however, the most recent experiments and ab initio calculations [584-587] suggest that at low dopant concentrations, Y is instead next near neighbor to the vacancy (Fig. 4.1(e)) and that lattice strain covers a fundamental role in ion diffusional paths. First-principles calculations [588, 589] indeed agree with the latter results, showing that the next-nearest-neighbor position for Y is energetically favored with respect to the nearest-neighbor position. The analyses by Li and coworkers [586, 587] further suggested that the Y atoms have a composition-independent eightfold coordination shell (like in a perfect fluorite structure) in the tetragonal lattice, thus supporting the hypothesis that dopant ions are, on average, at next-nearest-neighbor positions with respect to oxygen vacancies. This peculiar (reciprocal) configuration of oxygen vacancies and sub-valent Y ions has the effect of reducing the average coordination number of the zirconium atoms from 8 (i.e., as in the cubic structure) to values closer to 7 (i.e., similar to the monoclinic phase), thus stabilizing the tetragonal phase at room temperature. Taking these experimental considerations as a guiding light, Fabris et al. [246] have recently proposed a theoretical model consistently corroborating a number of experimental studies [590]. In this model, the dopant cations are located in a fluorite-like cation lattice at next-nearest-neighbor sites with respect to the oxygen vacancies and are considered not to take any active part in the stabilization mechanism. Accordingly, stabilization is dominated by the crystal distortions that occur around the oxygen vacancies. When the concentration of vacancies is low, a relatively large volume of crystal is left in the fluorite structure, and it undergoes the tetragonal distortion just as the stoichiometric material. This distortion involves the coordinated motion of all the oxygen sites, so that also the atoms neighboring to the defects are dragged along the tetragonal distortion of the oxygen sublattice by a displacement, d (Fig. 4.1(b)). When the concentration of defects is high, there is virtually no undistorted cubic region and every oxygen atom is either itself a neighbor of a vacant site or at least four of its six neighboring

oxygen atoms are. It follows that there is hardly a region in which the atomic environment could undergo any further distortion. Accordingly, the radial distortions around the vacancies dominate and the resulting atomic structure is cubic only when seen as an average over a large number of atoms, while its short-range atomic arrangement does have tetragonal symmetry (Fig. 4.1(f)). Besides reconciling many different (and apparently contradictory) experimental observations about zirconia stabilization, as offered through the years by many researchers from quite different viewpoints, the model proposed by Fabris et al. [246] is intriguing because it suggests that stabilization of the tetragonal zirconia polymorph at room temperature may be achieved, in theory, by “doping” zirconia crystals with oxygen vacancies only, the electronic and structural properties of (partly) stabilized zirconia being mainly controlled by structural disorder and by the local strain stored around the oxygen vacancies rather than by a direct effect of the cation dopant. Surface structures impart crucially important properties to the performance of a load-bearing biomaterial. Accordingly, experimental techniques and computational modeling have been extensively applied to investigate surface structures and properties of zirconia ceramics. One aspect particularly important here is the difference between surface and bulk stoichiometry. A large body of experimental and computational evidence has proved that Y in stabilized t-ZrO2 segregates to the surface. From the experimental side, X-ray photoelectron spectroscopy studies by Morterra et al. [591] indicated that Y segregation in t-ZrO2 is favored at low levels of yttria content (~2 mol.% Y2O3). Y surface segregation has also been confirmed by low energy ion scattering, which indicated that there is a high enrichment of yttrium on surface monolayers [592, 593]. A study by Bernasik of surface segregation in Y2O3-stabilized zirconia shows indeed a Y level at the surface in slight excess of the bulk concentration [594]. Moreover, an upper concentration limit of yttrium at the surface of Y2O3-stabilized zirconia has been observed by Zhu et al. [595], suggesting that the surface composition is independent of Y concentration in the bulk due to the segregation process. From the computational side, Ballabio et al. [596] have confirmed segregation by calculations for the main surfaces (111), (110) and (100) of an Y2O3-stabilized

crystal. From ab initio studies by Eicher and Kresse [597], it is found that the (101) surface of Y2O3-stabilized t-ZrO2 becomes more stable if the dopants are located close to the surface, which indicates a driving force for Y surface segregation. In contrast, no significant driving force for Y surface segregation was found for the (001) surface, thus suggesting that stoichiometric oxygen-terminated surfaces are more stable than Zr or O-O terminations. This latter statement might indeed be true independent of the specific orientation of the surface. Since surface orientation is found to cause significant variation in the local concentration of yttrium, all the properties that are a function of defect concentration and distribution (e.g., phase stability) must also be surface dependent. In other words, in polycrystalline materials, surface segregation and, thus, charge compensating oxygen anion vacancies are likely to experience substantial gradients on and below the material surface, with surface chemistry and stoichiometry effects dominating the statistical presence of active sites for the formation of metastable nuclei in vivo. According to the above considerations, it is clear that the stoichiometry of zirconia (especially at its free surface) plays a fundamental role in phase stability and, thus, in the overall performance of the material in bearing components. While this peculiarity has indeed offered wide material design opportunities in many different fields of application, it also forewarns technologists wanting to use zirconia as a structural biomaterial. The human body certainly represents a puzzling chemical environment, in which oxygen vacancies can be easily annihilated and/or cations replaced, especially due to the combined effects of chemical species and high strain gradients at the bearing surfaces. It should be thus very surprising if highly “susceptible” ceramic compounds, as the family of partly stabilized zirconia materials, could preserve unchanged for long time the same stoichiometry given them at the time of material fabrication. In the previous chapter, we have discussed in some detail how even an extremely stable oxide like alumina might undergo substantial stoichiometric changes in the severe tribochemical environment of a human hip joint. We can thus be sure that any (tetragonal) zirconia material embedded in the human body will necessarily change its chemical (and consequently its crystallographic) structure, the change being

accelerated by the presence of mechanical stresses and thus preferentially starting from the bearing surfaces. Polymorphic transformation will occur; it is just a matter of time. Pioneering examples of developing Y-TZP as a structural material and of considering it as a suitable alternative to alumina ceramics are presented in early papers by Rieth et al. [598] and Christel [599], respectively. A massive industrial development followed since the middle of the 80s, in the attempt of manufacturing femoral heads capable of overcoming the intrinsically limited mechanical properties of alumina ceramics. Explicit evidence of this vast industrial development in Japan remains attested in a paper by Tateishi and Yunoki from the early 90s [600]. Outside Japan, an earliest development focused on magnesia-partially stabilized zirconia materials. However, this material showed a peculiar structure (quite different from Y-TZP), with the tetragonal phase taking the morphology of small acicular precipitates embedded in a matrix of large equiaxed grains of cubic polymorph (i.e., typically 40~50 μm in diameter). As the presence of such a relatively “soft” matrix (i.e., as compared to Y-TZP and to the quite hard alumina biomaterials) was found to negatively influence the wear properties, most of the successive developments were focused on Y-TZP, a ceramic alloy with a more “standard” microstructure. Y-TZP consists of a homogeneous network of equiaxed submicron-sized grains that can be partly stabilized in their tetragonal phase with grain sizes easily confined to <0.4 μm. Since the 90s, Y-TZP (with 3 mol.% Y2O3; henceforth, 3Y-TZP) has been considered to be the standard material for clinical applications [601]. Note that using 3Y-TZP indeed relieves several shortcomings related to the use of alumina bioceramics, as discussed in the previous chapter of this book; for example, the higher crystallographic symmetry of the tetragonal structure as compared to the asymmetric hexagonal (corundum) structure of alumina minimizes the grain-boundary residual stresses arising from thermal expansion anisotropy, with direct beneficial aspects for the macroscopic material strength. Furthermore, fully dense 3Y-TZP ceramics can be obtained at sintering temperatures several hundredths of degrees centigrade lower than alumina. It follows, as explicitly shown in a later section, that through introducing a pre-sintering (low-temperature) HIPing cycle in the manufacturing process fully dense samples with a

fine and homogeneous microstructure can be promptly obtained. Their grain size is typically in the order of few hundredths of nanometers with a quite sharp grain-diameter histogram, a task that has definitely proved impossible for sintered monolithic aluminas (cf. histograms in Fig. 3.27). However, it should be kept in mind that the fraction of tetragonal phase retained at room temperature will strongly depend on grain size and on the uniformity in concen-tration of the yttria stabilizing oxide throughout the microstructural network. In other words, the mechanical properties of 3Y-TZP ceramics are strongly affected by manufacturing accuracy and rely on a delicate equilibrium among several microstructural parameters (e.g., dopant dispersion, sintering temperature, etc.). The effect of slight fluctuations in such parameters, as we will see in a forthcoming section of this chapter, can inflict a blow to the ultimate performance of zirconia as a biomaterial, even if fluctuations in processing (and microstructural) parameters are confined to ranges considered to be standard for alumina ceramics. As previously discussed in Section 1.3, the worst (historical) example of ceramic implant failures has certainly been the massive in vivo fracture of 3Y-TZP femoral heads in the year 2001. However, even leaving aside such a drastic situation, in vivo phase transformation may also involve detrimental effects on the wear behavior of the polyethylene sliding-counterpart. Serious concerns arose in the past on the effect of steam sterilization of 3Y-TZP femoral heads, which was found to have an impact on surface roughness due to the tetragonal-to-monoclinic phase transition. In 1996 [602], in classifying the causes of a number of hip revision surgeries due to osteolysis, surface degradation of retrieved zirconia femoral heads was found to coincide with high wear of UHMWPE sockets. Remarkably, it was found that the femoral heads showing the strongest surface degradation were those very same heads that, although supplied in their sterile status to the hospital, were re-sterilized in steam before implantation. On the positive side, a major discovery the long years of material development has been the effect of a small addition of Al2O3 in sintering 3Y-TZP, as comprehensively summarized in recent papers by Guo [603, 604]. There is a double benefit in adding a small amount (typically a fraction of 0.25 wt.%) of Al2O3 to 3Y-TZP: (i) an easier densification (i.e., meaning a lower sintering temperature and a finer microstructure [605, 606]); and, (ii) a better environmental

resistance in water-vapor environment [607]. However, it might be difficult to solve problems related to the heterogeneity of the small alumina quantities mixed with the 3Y-TZP matrix, due to the strong tendency to agglomeration of raw nanometersized powders [608]. It has been then suggested that uniformity in the incorporation of a minor constituent into the matrix can be ameliorated using a coating processing on the raw powder, by means of a colloidal processing or a sol-gel route, a solution coating or a precipitating coating [609, 610]. Although yet not applied to biomedical components, several other methods have also been proposed to avoid low-temperature thermal degradation of 3Y-TZP in hydrothermal environment. Watanabe et al. [611] found that below a critical grain size, no polymorphic transformation occurs after ageing in air up to 103 h. These researchers estimated the critical grain size being between 0.2 and 0.6 μm for yttria contents between 2 to 5 mol.%. Nitridation [612] has also been proposed as an effective method to enhance the resistance to hydrothermal degradation. In an early study, Hernandez et al. [613] observed that bulk-alloying 3Y-TZP with more than 2 mol.% of CeO2 can significantly increase the hydrothermal resistance. However, such process might lead to a relatively large increase in grain size (i.e., in the order of 2~3 μm), with subsequent reductions in strength and hardness. Sato et al. [614] made another interesting discovery as early as 1986. Upon calcining a sintered body of 3YTZP on a CeO2 powder bed at 1400°C for durations up to 10 h, these researchers showed that about 10 mol.% Ce could diffuse deeply into the material. Such heat-treatment greatly enhanced the thermal stability in air of the material, with exposures at 200°C up to 100 h leading to no monoclinic phase formation. Note that this actually was one of the earliest evidence that surface modification can enhance the resistance to hydrothermal degradation of 3Y-TZP without affecting its bulk mechanical properties. In the remainder of this chapter, we will proceed into the details of the complex behavior of zirconia-based biomaterials, with exploring some physical and mechanical aspects of their structural performances in vitro and in vivo. From a closer look, we will be able to judge better about the actual potentiality of this intriguing family of materials. While it remains a clear fact that zirconia ceramic implants have gone through a quite controversial history regarding their phase metastability, degradation in water

lubricants in simulation studies, and influence on friction and wear phenomena, it is also true that zirconia is a unique ceramic material with its potential for an extremely high toughness. A better correlation between material properties and biological performance appears to definitely be the key to pursue successful designs of improved bioceramics. In this context, it is quite conceivable that some unknown family of zirconia ceramics could be newly discovered, which possess peculiar microstructural features leading to so far unexplored mechanical and chemical properties. 4.2  The Controversial Effects of Polymorphic 

TransformationIn the previous section, we have discussed about the possibility of obtaining, at room temperature, a zirconia polycrystal in the metastable tetragonal form through the addition to the raw powder of an appropriate fraction of oxides with cations sub-valent to Zr4+ (e.g., Y3+, Ce3+, Mg2+, Ca2+, etc.). The obtained t-ZrO2alloy will eventually transform into its more thermodynamically stable monoclinic phase under the tensile stress field generated around the tip of a propagating crack. This phenomenon is known as stress-induced transformation toughening (cf. Fig. 2.5(b)). Due to the relatively large volume expansion and shear strain (≈3~5% and ≈7%, respectively) associated to the transformed lattice and the consequent compressive stress field generated upon constraint by the untransformed surrounding material, the local crack tip stress intensity is reduced and so the driving force for crack propagation (i.e., with an apparent increase of the effective toughness of the material). Transformation toughening is thus the result of an extrinsic toughening mechanism associated with the increase in critical stress intensity factor needed for propagating a crack. It is intuitive that effectiveness in polymorphic transformation requires the retention of relatively large fractions of metastable tetragonal phase at room temperature after manufacturing. Therefore, polymorphic transformability is generally regarded as a direct measure of the ability to release a crack-tip stress field [615, 616]. Unlike bone, the majority of synthetic ceramic materials possess

Figure 4.2 (a) SEM micrographs of the crack-tip neighborhood in polycrystalline Al2O3 (the arrow locates the location of the crack tip); (b) map of the trace of principal stress tensor collected by Raman piezo-spectroscopy from the same area in (a).poor resistance to fracture propagation. They thus need to be toughened by extrinsic mechanisms operating both in the crack wake and ahead of the crack tip. Such toughening effects give an assurance of reliability after implantation in the human body. Macroscopic fracture mechanics studies have provided quantitative characterizations of the actual resistance of bone to fracture in terms of critical stress intensity factor and critical strain energy release rate as measured at the onset of crack initiation [617, 618]. This approach has also been coupled with characterizations of toughness as a function of crack length (rising R-curve behavior), which is useful to quantify the toughening contribution arising from microscopic mechanisms occurring behind the advancing crack front [277, 278, 619, 620]. This set of characterizations has allowed scientists for an improved understanding of the fracture phenomenon in biomaterials. The same methods can be used to assess the actual potentiality of transformation toughening mechanisms in zirconia, thus enabling to take one step forward from the viewpoint of developing highly tough synthetic materials. From the perspective of microscopic characterizations, information of the stress patterns developed around the crack path in (metastable) tetragonal zirconia polycrystals leads to a better understanding of the relationships between toughness and polymorphic transformation zone. Microscopic fracture studies

employing a Raman microprobe might enable direct investigations on how a macroscopically applied external load engenders trans-formation at the microscopic level, thus linking the macroscopic toughening effect to microstructural parameters [621, 622]. The stress intensification at the tip of a propagating crack can be also directly measured by the COD method in the electron microscope (cf. Section 3.5) and the partial stress release linked to the occurrence of a crack-shielding mechanism. It should be noted that the use of a confocal probe enables to minimize in-depth averaging of the Raman signal, so that relatively sharp stress maps can be obtained. In crack-tip Raman experiments, the confocal probe can also be shifted below the free surface of the sample (by about 10 μm) in order to minimize disturbing effects arising from surface roughness and stress-releasing effects associated with the presence of the free surface. In this context, it is interesting to compare the crack-tip stress field developed (at the threshold for crack propagation) in a brittle biomaterial as synthetic polycrystalline alumina (cf. scanning electron micrograph and stress map in Figs. 4.2(a) and (b), respectively) with that developed in a toughened biomaterial as partially stabilized zirconia. The stress field in alumina can be visualized by monitoring the stress response (i.e., the so-called piezo-spectroscopic (PS) behavior) of the 418 cm-1 Raman band of alumina [623]. In finely grained biomedical alumina, no toughening mechanisms can be found around the crack tip and thus no stress relaxation effect is recorded in the microscopic crack-tip stress map. Under such micromechanical conditions, the crack-tip stress field preserves both magnitude and morphological symmetry characteristics, as expected in linear elastic materials [624]. In other words, no crack-tip toughening effect can be operative in arresting or delaying crack propagation, with the material fracturing with no microstructural “reaction” to the advancing crack. On the other hand, partially stabilized tetragonal zirconia is among the toughest synthetic biomaterials. As already mentioned, the improvement in mechanical properties, including toughness, strength and Weibull modulus, has been theoretically related to stress-induced phase transformation from the tetragonal to the monoclinic polymorph, which takes place in the neighborhood of a propagating crack [615, 616, 625]. Raman spectroscopy can visualize the occurrence of transformation

toughening mechanisms through providing quantitative microscopic information on the effect of a polymorphic transformation on crack-tip stress patterns developed ahead of an advancing crack. Figures 4.3(a)~(c) show the monoclinic transformation field in the neighborhood of the crack tip and, in the same zone, the stress field developed in the tetragonal and in the monoclinic phase, respectively. The corresponding equilibrium stress field acting on the crack faces is given in Fig. 4.3(d). This latter stress field, compressive in nature, represents the stress field computed as the average (weighted by the respective volume fractions) of the stress fields respectively stored in the freshly transformed monoclinic phase and in the residual tetragonal phase. Stress fields can be evaluated by exploiting the PS behavior of selected Raman bands of monoclinic and tetragonal zirconia polymorphs [117, 626, 627]. As already mentioned, significant volume expansion of the crystallographic lattice occurs upon stress-induced tetragonal-to-monoclinic transformation, which in turn induces the observed compressive stress field shielding the crack mouth. In other words, partly stabilized zirconia materials are capable to release the tensile crack-tip stress through the development of a selfinduced highly compressive stress field arising from polymorphic transformation. Such an intrinsic material “reaction” against fracture propagation can be thoroughly captured on the microscopic scale by Raman microprobe spectroscopy. Besides the importance of a stress visualization procedure from the basic materials science viewpoint, the quantitative knowledge of the exact amount of polymorphic transformation and the related residual stress magnitude is a fundamental step in improving the microstructural design of zirconia ceramics. It also helps materials scientists and technologists in correctly understanding the difference in mechanical properties among zirconia materials designed to achieve different microstructures through different chemical compositions. From this perspective, Raman spectroscopic evaluations can be considered a complementary tool to the set of macroscopic fracture mechanics characterizations, the latter locating the critical stress intensity factor at the onset for crack initiation (as well as the rising R-curve behavior), while the former visualizing the physics behind such mechanistic results.