Biomacromolecules 2005, 6, 942-947
942
Novel Route to Fatigue-Resistant Fully Sintered Ultrahigh Molecular Weight Polyethylene for Knee Prosthesis S. Rastogi,* L. Kurelec, D. Lippits, J. Cuijpers, M. Wimmer, and P. J. Lemstra Department of Chemical Engineering, Dutch Polymer Institute/Eindhoven University of Technology, Den Dolech 2, P.O. Box 513, 5600 MB Eindhoven, The Netherlands Received October 8, 2004; Revised Manuscript Received December 9, 2004
The role of entanglements in obtaining a homogeneous product of ultrahigh molecular weight polyethylene (UHMW-PE) has been explored. Studies performed in this report show that a disentangled state before melting is a prerequisite to obtain homogeneous products of an intractable polymer like UHMW-PE. The disentangled state is obtained directly from the reactor by controlling the polymerization conditions or in the solid state when there is enhanced chain mobility along the c axis of a unit cell. The disentangled state is maintained in the melt over a period of time, invoking implications in polymer rheology. This approach is applicable to polymers in general. The homogeneous fully sintered UHMW-PE, obtained for the first time, shows a considerable decrease in oxygen permeability and an increase in toughness and fatigue resistance. Such homogeneous products of UHMW-PE are beneficial in highly demanding applications, especially in knee prosthesis, where the polymer is used as an inlay between the human bone and a metal or ceramic part, which slides against the polyethylene component during normal gait. Introduction One of the main factors contributing to the popularity of synthetic polymers in every day life is their ease of processing into complex shapes. Processing routes conventionally applied in the polymer industry are injection molding, extrusion, film blowing, and blow molding.1 All these processes start from a melt, and the molten state is mostly affected by the molar mass. This is given by the universal relationship between zero shear viscosity (η0) and molar mass (Mw), which for polymers having high molar mass reads as follows:2 log η0 ) C + 3.4 log Mw, where C is a constant (1) Such a strong dependence between η0 and Mw is associated to the number of entanglements, which increases with Mw. This imposes a restriction on the flow ability of the melt. In such a highly entangled melt, the motion of a chain is extremely slow. This is postulated by the reptation model,3,4 in which a chain is considered to move wormlike through a virtual tube, formed by the neighboring chains. The time needed for a chain to renew its own tube, that is, to change its position within the melt and relax, is also highly dependent on molar mass (τ0 ∼ Mw3.4). These two fundamental restrictions, related to viscosity and motion of chains within the melt, make high molar mass polymers rather intractable via conventional processing routes. On the other hand, final properties such as toughness, strength, and wear increase with molar mass. Ultrahigh molecular weight polyethylene (UHMW-PE) is currently processed via compression molding or ram extru* Author for correspondence and requests for materials. E-mail: s.rastogi@ tue.nl.
sion into simple shapes such as rods, plates, or sheets, which are subsequently machined into the desired products.5 It has been found that all processed products of UHMW-PE possess residues of the original powder particles usually referred to as grain boundaries or fusion defects.6 These flaws in the material are a consequence of the long reptation time needed for the molecular chain to cross from one powder particle to another.3,4 This means that to improve the processability of this material, it is necessary to decrease the reptation time by reducing the number of entanglements per chain in the melt. It is possible to obtain a less entangled system directly in the reactor, by controlling the polymerization conditions. Several nascent powders have been recognized to be drawable at temperatures below the melting temperature of polyethylene. This drawability was associated to a reduced number of entanglements.7,8 By further controlled chemistry, that is, decreasing number of active sites or using a single site catalyst,9 and/or decreasing polymerization temperature lower than the crystallization temperature, it is ultimately possible to make “monomolecular crystals”, that is, a single chain forming a single crystal. Materials and Sample Preparation For our studies, three different grades of polymers are examined: grade A is a commercial polymer synthesized with a highly active Ziegler-Natta catalyst (Z-N), used for medical applications. Grade B is a laboratory scale polymer synthesized at low temperature with a Z-N catalyst. Because of the low polymerization temperature the crystallization rate of this grade is higher than the polymerization rate of grade A, and, consequently, a less entangled nascent powder is
10.1021/bm0493638 CCC: $30.25 © 2005 American Chemical Society Published on Web 01/29/2005
Biomacromolecules, Vol. 6, No. 2, 2005 943
UHMW-PE for Knee Prosthesis Table 1
commercial, Z-N (grade A) laboratory, Z-N (grade B) metallocene (grade C)
Mw [g mol-1]
Mw/Mn
Tsynthesis (°C)
4.54 × 106 3.6 × 106 3.6 × 106
10.0 5.6 3.0
60-80 50 25
obtained. Grade C is a sample synthesized with a homogeneous, single-site, metallocene catalyst at an even lower temperature. Because the active sites in a metallocene catalyst are much further apart compared to those in the Z-N catalyst, this grade has a much lower number of entanglements per unit chain. The molecular characteristics of the three grades are summarized in Table 1.To avoid oxidation of the nascent powder 1% of anti-oxidant Irganox B215 was dissolved in acetone and dispersed over nascent UHMWPE powder. The acetone was evaporated. Samples were made by placing the nascent powders between aluminum foils and compacting them into films using a pressure of 200 bar and a temperature of 70 °C for 30 min, below the R-relaxation temperature to maintain the initial morphology. Small Strain Viscoelastic Dynamic Rheometry. Small strain dynamic viscoelastic rheometry was performed using an Ares 3LS-4A, Rheometrics, Inc., with a plate-plate geometry having a diameter of 25 mm. Samples having a diameter of 12 mm and a thickness of 1 mm were measured. All measurements were performed in a temperature range of 160-180 °C in a nitrogen environment to prevent degradation. The dynamic strain applied was 0.5%. At this strain level all the tested materials were fully viscoelastic. The annealing and heating steps were performed inside the rheometer. An oscilloscope was attached to the apparatus to verify that no slippage between the sample and the parallel plates had occurred. After the measurements the geometry of the samples was measured again. To verify that no oxidation to the samples has occurred infrared spectroscopy before and after the experiment was performed, using a BioRad FTS 6000 spectrometer. To check that no cross-linking had occurred, the samples were dissolved in xylene. Results and Discussion To investigate the entanglement density in the melt, rheological experiments were performed. It is possible to calculate the average molecular weight between the entanglements, 〈Me〉, from the elastic modulus of the rubber plateau:2 G ′plateau ) gNFRT/〈Me〉
(2)
where gN is a numerical factor and F is the density. To investigate the entanglement density of the different materials listed in Table 1, the samples were molten for 15 min at 160 °C and a frequency sweep was recorded in the melt at this temperature. Figure 1a shows that the plateau modulus of the commercial grade Z-N is approximately 2.0 MPa. This is comparable to the value of the rubber plateau reported in the literature.10 The laboratory Z-N grade shows a lower plateau modulus of 1.1 MPa and thereby a larger 〈Me〉, whereas the metallocene grade shows a plateau modulus of
0.6 MPa. The low plateau modulus of the metallocene sample suggests the largest 〈Me〉 of the three samples. At this stage it is essential to add that because the starting material of grade C is disentangled and, thus, is not in the thermodynamically nonequilibrium state, eq 2 in principle cannot be applied. It is also to be noted that no rheological equations reported in the literature to date are valid for such a nonequilibrium state. However, for present studies we will make use of this equation to “provide an estimation” on changes in molar mass between entanglements. These concepts and results in detail have been described elsewhere (see ref 12) in one of our manuscripts. Figure 1b shows that when the metallocene synthesized polymer is left under constant strain and fixed frequency (10 rad/s) and temperature (180 °C), the plateau modulus increases gradually with time from 0.6 to 2.0 MPa, suggesting a gradual decrease in 〈Me〉 due to reptation of the initially less entangled molecules. It is to be noted that the time required for the complete entanglement process could exceed more than 2 h which will be a combination of Rouse11 and reptation time.3 From here, it can be concluded that by adjusting the polymerization conditions the number of entanglements per unit chain can be controlled. Detailed studies on thermorheological behavior of metallocene-based polyethylenes have been performed by us12 and Wood-Adams and Costeux,13 and a new quantitative development in the reptation model has been proposed by Likhtman and McLeish.14 At this point we would also like to add that the observed buildup in modulus in metallocene samples is significantly different to the earlier reported immediate increase of modulus in the solution-crystallized films.10 Chains in the solution-crystallized samples, similar to those of metallocene grade, are disentangled. The disentangled crystals in the two samples provide high drawability in the solid state, close but below melting temperature. The reason for the differences in the modulus buildup, on melting of the two samples, exists in the major distinction in the morphology of the two samples that they acquire upon crystallization. Crystals in the metallocene samples do not possess regular registration, which is likely to be a cause for the normally reported high melting temperature (142 °C) on first heating,12 whereas crystals in solution-crystallized films are regularly stacked. These crystals melt at normal 135 °C. The regular registration of crystals in the solution-crystallized films provides enhanced chain mobility above the R-relaxation temperature, promoting the lamellar doubling process.15 This helps in immediate chain entanglement upon melting, whereas in nascent samples the absence of crystal sedimentation requires chain reptation for the formation of entanglements, which in rheology is seen as modulus buildup. Another route to disentangle the already entangled chains is via increasing chain mobility along the c axis of the unit cell.16 In the solid state of polyethylene, this is obtained via the hexagonal phase.17 Nascent powders, synthesized at temperatures below the melting temperature of polyethylene, consist of crystals having very small dimensions in thickness as well as lateral direction (10-20 nm).18-20 For such small folded chain crystals, the hexagonal phase is obtained at
944
Biomacromolecules, Vol. 6, No. 2, 2005
Rastogi et al.
Figure 1. (a) Plateau modulus of the three grades over frequency, at a fixed temperature of 160 °C. The constant dynamic strain applied on a compressed UHMW-PE plate of diameter 12 mm was 0.5%. Because of the elastic nature of the ultrahigh molecular weight of the polymer, the plateau modulus remains constant over the applied frequency. The decrease in the plateau modulus corresponds to an increase in 〈Me〉. These results are in accordance with the polymerization conditions of the three grades. (b) Low plateau modulus of the metallocene-synthesized polymer corresponding to the relatively low number of entanglements. The figure shows that the plateau modulus recorded at the constant strain (0.5%) and frequency (10 rad/s) at fixed temperature (180 °C) builds up gradually with time to a constant value of 2 MPa. The time needed for the buildup of the modulus is approximately 2 h. These results point to the ongoing reptation process from the initially less entangled to the thermodynamically entangled state. No change in the phase angle with time further confirms that the chemical nature of the chains does not alter with time. To avoid any chemical changes, antioxidant was added to the samples and the experiments were performed in a nitrogenrich environment. Infrared spectroscopy, dissolution methods, and intrinsic viscosity measurements were used to monitor any possible chemical changes.
pressures much lower than the equilibrium triple point21,22 (Q0; Figure 2). To have an effective disentanglement process, substantial crystal thickening is required in the hexagonal phase. This will pull a large part of an entangled chain into a crystal. The thickening process of nascent crystals in the hexagonal phase has been studied on two grades, grades A and B, of UHMW-PE synthesized by Z-N catalyst. Both samples were compacted at 1500 bar and subsequently heated isobarically at 2 °C/min up to the full transition of crystals into the hexagonal phase (Figure 2). To get an insight into how the hexagonal phase helps in the thickening process of the nascent powders, the heating was stopped before melting of the crystals. The phase transformations during compaction and heating were monitored in situ with the help of Raman
spectroscopy and/or X-ray, which is capable of distinguishing between different phases in polyethylene under pressure.23 Figure 2 shows transmission electron micrographs of the samples after the pressure-temperature cycle as indicated in the phase diagram. These micrographs reveal the ability of nascent crystals to thicken within the hexagonal phase, leading to a disentangled state, where the thickening rate is highly dependent on the initial entanglement density. For the grade possessing a lower initial entanglement density (grade B), even relatively short residence times within the hexagonal phase are sufficient for crystals to thicken up to a value of 100 nm. In contrast, for the initially highly entangled state (grade A) the same time within the hexagonal phase leads to a much lower crystal thickness of not higher
UHMW-PE for Knee Prosthesis
Figure 2. The full lines in the phase diagram represent the thermodynamic equilibrium position of the phase transitions for the extended chain crystals. The dotted lines correspond to the metastable region in the phase diagram, for folded chain crystals. The position of the meta-stable region depends on the crystal size, which changes in time with crystal thickening.22 The experiments depicted in this figure are performed in the meta-stable region of the hexagonal phase. The transmission electron micrographs (TEM) below the phase diagram show broad bands that represent polyethylene crystals viewed edge-on. The chain direction is perpendicular to the crystal length.21 The two Z-N grades, grades A and B, were heated (2 °C/ min) along the path shown in the phase diagram and were left to anneal within the meta-stable hexagonal phase for equal residence times. A comparison of the two micrographs shows that the crystals in grade B (R) thicken more than the crystals in grade A (β).
than 30 nm. Consequently, the disentanglement process of the initially highly entangled system requires much longer residence times within the hexagonal phase than in the case of the initially less entangled grade.
Biomacromolecules, Vol. 6, No. 2, 2005 945
Because of the entropically driven increase in the radius of gyration during melting, as shown for the solution crystallized samples, the disentangled sample would be a good prerequisite for successful sintering.24,25 This provides the instant enhanced chain mobility needed for crossing the grain boundaries. In our samples, this can be proven by the following experiment. Figure 3, micrograph R, shows that the sample (grade B) which has been thickened in the hexagonal phase along path 1 but not melted still exhibits grain boundaries. This indicates that the intervention of the hexagonal phase alone is not sufficient to obtain the fully fused texture of the sample. The full fusion of the original powder particles only occurs upon melting the disentangled sample, even along path 3 after reducing the pressure back to ambient (Figure 3, micrograph β). The same set of experiments has been performed also with the commercial grade of UHMW-PE (grade A), which following the same pressure-temperature paths does not fuse (Figure 3, micrograph γ). For this grade, compared to grade B, much longer annealing times within the hexagonal phase are needed to achieve the fully homogeneous texture. On the other hand, the sample for which the disentangled state is obtained during synthesis (grade C) sintering on melting can occur even without the prior intervention of the hexagonal phase, leading to full fusion of particles at pressures as low as 200 bar. As we are now able to obtain, for the first time, fully fused samples of UHMW-PE, extensive mechanical tests have been performed to understand the influence of grain boundaries. For example, in the knee prosthesis, the UHMW-PE inlay has to sustain the mechanical stresses during normal gait, a complex nature of cyclic loading.26,27 In the nonsintered UHMW-PE, crack formation and subsequent crack propagation play a dominant role in the failure of the polyethylene part.28,30 Because we are now able to make fully sintered UHMW-PE these problems could be overcome.
Figure 3. Micrograph R shows the presence of grain boundaries in the grade B sample that has gone through paths 1 and 2. The micrograph β shows that this sample becomes fully sintered on melting at ambient pressure, along path 3. On comparing parts R and β it is evident that the disentangled state obtained via the hexagonal phase promotes the complete fusion of the particles. The result of fusion of the commercial grade, grade A, after paths 1-3 is given in the micrograph γ. Because this sample does not disentangle on annealing in the hexagonal phase, it retains the grain boundaries. The conditions applied are the same as those discussed in Figure 2.
946
Biomacromolecules, Vol. 6, No. 2, 2005
Figure 4. Crack length [a in mm] versus number of cycles [N] for different samples (a). “Ref.” refers to the reference material, a medical UHMW-PE grade that is used in knee prosthesis. “HDPE” is highdensity polyethylene, and “Hex” refers to the obtained grain-boundaryfree material. Part b shows Paris-Erdogan plots that have been calculated from the crack propagation data.30 The measurements were performed for an equal crack propagation rate [da/dN] according to ASTM E647-93. The table indicates the parameters of the Paris regime calculated from the plots,30 where ∆Kint is related to the stress required for the crack to start to propagate. When comparing the results for the different polyethylene samples, the “Hex” sample exhibits the highest fatigue resistance, indicating that the absence of grain boundaries highly improves the fatigue resistance of the UHMW-PE.
The crack length versus number of cycles up to total failure is plotted in Figure 4. Under the applied loading conditions crack propagation in the fully sintered material is slower (more than two decades) compared to the conventional UHMW-PE. Hence, the presence of grain boundaries plays an important role in the fatigue resistance of UHMW-PE products. To strengthen this argument, the same type of measurements was performed on a lower molecular weight material like the high-density polyethylene sample (HDPE, having a molar mass of 60 000 and molecular weight distribution of 4). This sample does not feature any grain boundaries or any other fusion-related defects because of its lower molar mass. An interesting observation from this measurement is that HDPE has a higher fatigue resistance than the unfused UHMW-PE sample. At first glance, this result is rather striking because it is normally expected that the mechanical properties of ultrahigh molecular weight polymers will always exceed the properties of lower molecular weight polymers. However, these results confirm the fundamental postulate of fracture mechanics, that the fracture of the material is governed by the presence of internal defects.29,30 Therefore, the UHMW-PE possessing intrinsic defects will fail easier than the defect-free HDPE.
Rastogi et al.
Figure 5. Oxidation index as a function of depth in the ref (nonsintered product) and hex (fully sintered product) samples as measured at different times. The experiments reported in this figure have been performed following a method of Stanford et al. (O2 under 5 atm; 70 °C for different periods of time).31 The level of oxidation at different depths of the sample has been determined with the help of infrared which is a preferred method within the orthopaedic community.5 The samples were sterilized using γ irradiation and subsequently aged for different times in the presence of oxygen at 70 °C. Extent of oxidation has been measured by calculating the oxidation index, which is a ratio between the area under the carbonyl vibration (in the range 1680-1800 cm-1) and the area under 1370 cm-1, a mode sensitive for sample thickness.
Moreover, in fully sintered materials, because of the absence of grain boundaries, oxygen permeability decreases by nearly 50% and overall toughness increases by nearly 30%, before γ irradiation. These physical properties are dependent on Mw and Mw/Mn. Figure 5 shows a comparison of the oxygen permeability of UHMW-PE in the fully sintered product (made via the route described in this report) and that of the commercially available nonsintered product. The samples were subjected to an equal dose of γ irradiation (approximately 30 kGy), following the ASTM E691-92 standard. The decrease in oxygen permeability is of crucial relevance for the increase in the lifetime of knee prosthesis because fatigue-related failure of UHMW-PE is also dominated by oxidation that occurs secondary to irradiation, induced as a result of the formation of free radicals,31 whereas after irradiation no differences in toughness of the fully fused and not fused UHMW-PE samples are observed. Considering the knowledge of toughness in polymers this suggests that ultimately an initially disentangled material gets fully
UHMW-PE for Knee Prosthesis
entangled. Wear resistance of UHMW-PE under multidirectional motion relevant for the hip joint remains unchanged. In this report, taking UHMW-PE as an example, we have shown that ultimate physical and mechanical properties of a material can be obtained by adopting a multidisciplinary approach, combining chemistry, physics, and rheology. For example, by tailoring the initial morphology of the polymer, the disentangled state could be obtained, thus, avoiding the pressure requirement for making fully fused products. Creating the disentangled state in the melt is also beneficial to the polymer processing in general. Fundamental aspects of a nonequilibrium disentangled melt requires new approaches over the existing knowledge, where 〈Me〉 is considered an intrinsic property of a polymer. References and Notes (1) Cahn, R. W.; Haasen, P.; Kramer, E. J. Materials Science and Technology: Processing of Polymers; VCH: Weinheim, 1997; Vol. 18. (2) Ferry, J. D. Viscoelastic Properties of Polymers; John Wiley & Sons: Toronto, 1980. (3) de Gennes, P. G. J. Chem. Phys. 1971, 55, 572. (4) Doi, M.; Edwards, S. F. The Theory of Polymer Dynamics; Oxford Press: Oxford, 1986. (5) Kurtz, S. M.; Muratoglu, O. K.; Evans, M.; Edidin, A. A. Biomaterials 1999, 20, 1659. (6) Jenkins, H.; Keller, A. Macromol. Sci. Phys. B 1975, 11, 301. (7) Rotzinger, B. P.; Chanzy, H. D.; Smith, P. Polymer 1989, 30, 1814. (8) Kanamoto, T.; Okama, T.; Tanaka, K.; Takede, M.; Porter, R. S. Polymer 1987, 20, 1517. (9) Scheirs, J.; Kaminsky, W. Metallocene-based polyolefins: preparation, properties and technology; Wiley: Chichester, 2000. (10) Bastiaansen, C. W. M.; Meijer, H. E. H.; Lemstra, P. J. Polymer 1990, 31, 1435.
Biomacromolecules, Vol. 6, No. 2, 2005 947 (11) Rouse, P. E., Jr. J. Chem. Phys. 1953, 21, 1272. (12) Rastogi, S.; Lippits, D.; Peters, G. W. M.; Meijer, H. E. H. Macromolecules 2005, in press. (13) Wood-Adams, P.; Costeux, S. Macromolecules 2001, 34, 6281. (14) Likhtman, A. E.; McLeish, T. C. B. Macromolecules 2002, 35, 6332. (15) Rastogi, S.; Spoelstra, A. B.; Goossens, J. G. P.; Lemstra, P. J. Macromolecules 1997, 30, 7880. (16) Maxwell, A. S.; Unwin, A. P.; Ward, I. M. Polymer 1996, 37, 3282. (17) Basset, D. C.; Khaifa, A.; Turner, B. Nature (London) 1972, 239, 109; Nature (London) 1972, 240, 146. (18) Tervoort-Engelen, Y. M. T.; Lemstra, P. J. Polym. Commun. 1991, 32, 345. (19) Phillips, R. A. J. J. Polym. Sci., Part B: Polym. Phys. 1998, 36, 495. (20) Corbeij-Kurelec, L. Ph.D. Thesis, Eindhoven University of Technology, Eindhoven, The Netherlands, 2001 (ISBN90-386-3032-8; http:// alexandria.tue.nl/extra2/200113706.pdf). (21) Keller, A.; Hikosaka, M.; Rastogi, S.; Toda, A.; Barham, P.; Gooldbeck-Wood, G. J. Mater. Sci. 1994, 29, 2579; Philos. Trans. R. Soc. London, Ser. A 1994, 348, 3. (22) Rastogi, S.; Kurelec, L.; Lemstra, P. J. Macromolecules 1998, 31, 5022. (23) Kurelec, L.; Rastogi, S.; Meier, R. J.; Lemstra, P. J. Macromolecules 2000, 33, 5593. (24) Barham, P.; Sadler, D. M. Polymer 1991, 32, 393. (25) de Gennes, P. G. C. R. Acad. Sci., Ser. IIb 1995, 321, 365. (26) Hood, R. W.; Wright, T. M.; Burstein, A. H. J. Biomed. Mater. Res. 1983, 17, 829. (27) Landy, M. M.; Walker, P. A. J. Arthroplasty 1988, 3, S73. (28) Bartel, D. L.; Bicknell, V. L.; Wright, T. M. J. Bone Jt. Surg. 1986, 26A, 1041. (29) Anderson, T. L. Fracture Mechanics-Fundamentals and Applications; CRC Press, Inc.: Boston, 1991. (30) Paris, P.; Erdogan, F. Trans. ASME 1963, 528. (31) Sanford, W. M.; Saum, K. A. Accelerated oxidative ageing testing of UHMW-PE. Trans. 41st Orthop. Res. Soc. 1995, 20, 119.
BM0493638
