Biomaterids

@ 1998

18 (1997)

1659-1663

Published by Elsevier Science Limited Printed in Great Britain. All rights reserved
014%9612/98/519.00

ELSEVIER

PII:

SO142-9612

(97)

00124-5

Validation of a small punch testing technique to characterize the mechanical behaviour of ultra-highmolecular-weight polyethylene
Steven M. Kurtz, Jude R. Foulds, Charles W. Jewett, Sanjeev Srivastav and Avram A. Edidin*
Failure Analysis Associates, Inc., 2300 Chestnut Street, Suite 150, Philadelphia, Corporate, Allendale, NJ, USA PA 19103, USA; Osteonics, R&D

The small punch or miniaturized disc bend test has been used successfully to characterize the ductility and fracture resistance of metals and ceramics with specimens measuring 0.5 mm in thickness. This study was performed to demonstrate the feasibility of performing small punch tests on implant grade ultra-high-molecular-weight polyethylene (UHMWPE). Large-deformation finite element simulations were developed and validated to explore the hypothesis that the macroscopic constitutive behaviour of UHMWPE may be inferred from a miniature specimen testing technique which can be used to characterize the ductility and work to failure for UHMWPE. The load-displacement curve was insensitive to cyclic preconditioning of the test specimen and only mildly sensitive to the loading rate. Furthermore, the initial slope of the small punch load-displacement curve was used to determine the elastic modulus of the UHMWPE with the help of the inverse finite element method. The ultimate goal of this research is to develop the capability to perform local measurements of material tensile and static fracture properties in as-manufactured, as-sterilized and as-retrieved UHMWPE components. 0 1998 Published by Elsevier Science Limited. All rights reserved
Keywords: element Ultra-high-molecular-weight analysis
12 May 1997; accepted 30 June 1997

polyethylene,

mechanical

properties,

small

punch

test, finite

Received

For the past decade, the small punch or miniaturized disc bend test has been used successfully to characterize the duct.ility and fracture resistance of metals and ceramics with specimens measuring only 0.5 mm in thickness-3. Development of the small punch test for metall:ic materials has been driven by the need to measure in-service degradation of mechanical properties with a limited volume of available material. By virtue of its small specimen size, the test also provides a convenient means of characterizing material at spec:ific locations in a component or structure. By concurrently performing small punch tests and static fmcture tests, researchers have empirically correlated small punch mechanical behaviour with fracture toughness in metal and ceramic specimens (Xi, for brittle materials or /ICfor ductile materials23). The primary disadvantage of this empirical approach is that a large volume of bulk tensile and fracture data is required for a given material
in order to make reliable small punch test resu1t.s. engineering predictions from

Correspondence to Dr S. .Kurtz. Tel: 001 215 751 1661; fax: 001 215 7510660; fail.com

e-mail: skurtz@-

An alternate interpretation of the small punch test, developed by Foulds et a1.l uses the inverse finite element method to infer conventional tensile stressstrain properties based upon optimal matching of the experimental and simulated small punch loaddisplacement curve. The as-determined tensile stressstrain behaviour is then used to compute, by the finite element method, the local strain energy density accumulated to initiate cracking in the small punch specimen. This critical strain energy density, considered a material fracture initiation criterion, is analytically (by simulation) applied to the standard plane strain compact tension specimen crack tip to obtain the conventional fracture toughness, JrC (and I&). Tensile and fracture properties measured using this approach have been shown to be reasonably accurate for a wide range of metals, but have yet to be explored for polymers, such as ultra-high-molecularweight polyethylene (UHMWPE). Although UHMWPE has been used successfully in total joint replacements during the last three decades, there is increasing concern in the orthopaedic community that gamma radiation sterilization ma limit the long-term wear performance of the polymer4 -!I . As gamma-irradiated UHMWPE components age on
1659
Biomaterials 1997, Vol. 18 No. 24

1660

Small

punch

testing

of UHMWPE:

S. M. Kurtz et al.

the shelf or in the body, the physical properties of the polymer (such as density and crystallinity) degrade inhomogenously with depth6-a. The density of UHMWPE has been related to mechanical properties (such as elastic modulus or yield stress)g and to the shape of the true stress-strain curve in uniaxial tension by empirical relationships that were obtained from bulk mechanical tests. Therefore, it has been possible to infer that retrieved components, for which one can measure inhomogeneous density distributions also have inhomogeneous through the thickness, mechanical properties through the thickness. Miniature specimen testing provides an opportunity to directly measure properties in inhomogeneous UHMWPE components. For example, Collier et al5 measured the ultimate elongation and ultimate tensile miniature tensile of UHMWPE using strength specimens (0.2 mm thick) manufactured parallel to the surface of shelf-aged tibia1 trays and bar stock material. An order of magnitude reduction in ductility was observed between specimens prepared from a degraded subsurface region (characterized by a white band upon transverse microtoming) and specimens prepared from undegraded stock material. However, it remains measured from small unclear how properties specimens of UHMWPE may be effectively compared with macroscopic or bulk material properties. Consequently, the primary goal of the present study was to demonstrate the feasibility of performing small punch tests on specimens manufactured from implant grade UHMWPE. The secondary objective was to quantify the sensitivity of the small punch mechanical behaviour to changes in test conditions, such as loading rate. Large-deformation finite element simulations were developed and validated to explore the hypothesis that the macroscopic constitutive behaviour of UHMWPE may be inferred from micromechanical tests, Ultimately, the small punch testing technique in conjunction with the inverse finite element method may provide local, accurate estimates of both conventionally measured tensile and static fracture properties for UHMWPE components,

Alignment Asseebly Pin t T

3.8 mm dia. -_)I It

Alignment Pin Assembly Bolt t t

Guide

2.5 mm --) dia.

-L I/

Specimen:
6.4 mm dia. x 0.5 mm thick

Schematic specimen, the testing head punch.

Figure 1

of the disc-shaped small punch guide and die, and the hemispherical

MATERIALS AND METHODS Experimental

of 6.4mm Disc-shaped small punch specimens diameter and 0.5mm thickness were machined from a 64-mm-thick sheet of compression-moulded GUR4120HP resin (PolyHi Solidur, Fort Wayne, IN, USA), having a reported molecular weight of 2.0 to 2.5 million g mol-I. Small punch specimens were carefully machined from the stock material to avoid phase changes near the surface. Characterization of the UHMWPE physical properties using the density gradient column technique (DGC) and differential scanning calorimetry (DSC) revealed that the density of the virgin material was 0.934gcme3 and that the crystallinity was 51%. No traces of oxidation were observed at the centre of the machined specimen, as determined by Fourier transform infrared spectroscopy (FTIR). Since gamma sterilization and degradation were expected to substantially reduce the ductility, virgin material was chosen for this study to provide
Biomaterials 1997, Vol. 18 No. 24

the most-ductile small punch testing scenario for implant grade UHMWPE. The disc-shaped specimens were tested in bending by indentation with a custom-built, hemispherical head punch, as described previously (Figure ~)l. Specimens were tested at constant punch displacement rates of 0.25 mm min-l (number of specimens, R = l), 0.5 mm min-l (n = 5), 2.5mmmiri (n = 4) and 5.0 mm min-l (n = 4). During the testing, the punch load and displacement were digitally recorded, while the back (bulged) surface of the specimen was videotaped via a borescope. Selected specimens were gold-coated (2OOA thickness) and examined at x18 and higher magnifications using a scanning electron microscope at 10 kV to characterize the specimen before and after testing. Initial tests showed that the linear (elastic) portion of the load-displacement curve occurred at displacements of less than 0.064mm and at loads of less than 3N. Consequently, four additional tests were performed at a rate of 0.5 mmmin- using a low range (4.5N) on an 89 N capacity load cell. An extensometer was attached to the device for measurement of displacement (rather than machine actuator displacement, i.e. stroke, used earlier) to more accurately characterize the initial, linear portion of the load-displacement curve. During these tests, the samples were cyclically loaded 10 times in a range from approximately 0.2 to 2.7 N to demonstrate the stabilization (saturation) of the slope of the initial linear portion of the curve. After cyclic preconditioning, the range of the load cell was increased to 89 N and the samples were tested to failure at a rate of 0.5 mm min-I. The surface of the specimen during these low-amplitude cyclic loading tests was not videotaped. The effect of changing test conditions (e.g. loading rate, cyclic preconditioning) on features of the loaddisplacement curve, such as mean peak load, was determined by two-sample t-tests assuming unequal variance. A P-value of 0.05 was used for statistical significance.

Analytical
Two-dimensional (axisymmetric) finite element models were developed to aid in interpreting the experimental

Small punch testing of UHMWPE: S. M. Kurtz et a/.

1661 90

small punch tests results. The large-deformation structural problem was solved with NIKE2D (Lawrence Livermore National Laboratory, Livermore, CA, USA) elasticity using isotropic and rate-independent isotropic plasticity to model the mechanical response of UHMWPE to failure, based on previous bulk mechanical testing*, The plasticity model in NIKEBD was validated for UHMWPE by simulating a tension test to failure, which for bulk GUR4120 specimens were previously shown to occur at nominal (engineering) strains of 421%11. The axisymmetric: finite element model of the small punch test consisted of 1602 four-noded quadrilateral elements. Sliding contact between the specimen, punch head, punch guide and receiving die was modelled with slidelines using the penalty stiffness method12 and a coefficient of friction of 0.1 was used between the UHMWPE and the polished surfaces13. contacting steel The polyethylene specimen was modelled with 12 elements through the thickness, and zoning studies were performed to verify spatial and temporal convergence of the contact solutions. The resultant load1 acting on the punch head, in conjunction with the prescribed displacement, was used to compile a load-displacement curve for the small punch simulation. The elastic modulus for the simulated UHMWPE specimen was then parametrically varied between 255 and 102lMPa. For each of the parametric simulations, the initial slope of the loaddisplacement curve, 1:, was computed up to O.O64mm, which during preliminary experiments was observed to correspond with linear elastic behaviour. A linear least-squares correlation was then sought between the initial stifmess of the simulated load-displacement curve and elastic modlulus: k=AE (1)

I,,,,,,,,,,,,,,,,.

80 _

UHMWPE Ultimate Load 81 Displacement -

60 z u $ a 50 40 30

Punch Displacement

(mm)

Figure 2 Features of a typical load-displacement curve for the small punch test performed to failure on a UHMWPE specimen. failed catastrophically at an easily distinguishable, abrupt load-drop point on the load-displacement curve. Videotapes of the small punch tests further showed that the specimens underwent easily discernible changes in opacity during the test, consistent with phase changes in the crystalline structure of the polymer during large plastic deformations. The scanning electron micrographs of tested specimens provided additional evidence that the specimens were reproducible, large-scale plastic subjected to deformations (Figure 3). Micromachining lines were evident in the flat, as-manufactured specimens and were homogeneously stretched as the test progressed. However, the machining lines were obliterated during the large deformations culminating in specimen failure and probably did not affect the outcome of the small punch test.

where k is the initial stiffness, A is a constant coefficient, and E is the elastic modulus. Equation 1 was inverted to predict elastic modulus for GUR4120 based on the initial stiffness measured during the small punch test. During previous bulk mechanical testing of GUR412014, the elastic modulus was found to be 85l f 33 MPa (six specimens were tested in compression). A two-sample t-test assuming unequal variance wals used to compare the elastic modulus from the bulk mechanical test with the elastic modulus predicted by Equation 1.

RESULTS Experimental
The load-displacement behaviour of the small punch test specimen displayed distinctive features (Figure z), such as a peak load early during the test, followed by a comparatively long, plastic membrane stretching phase, analogous to the stretching of a bulk tensile specimen. The load and displacement at failure, as well as the energy or work to failure, provided additional quantitative measures of the specimen ductility and fracture resistance (Figure 2). Synchronization of the videotape and load-displacement curves demonstrated that the small punch test specimen

Scanning electron micrographs of a UHMWPE small punch specimen tested to failure (x18). Note the obliteration of machining lines near the summit of the specimen. Biomatetials
1997,

Figure 3

Vol.

18 No. 24

1662

Small

punch

testing

of UHMWPE:

S. M.

Kurtz et al.

90
80 70 60 5 i A 50 40 30 20 10 0 0 1 2 3 4 5 Rate = 0.5 mm/min

UHMWPE 2.5 - Rate = 0.5 mmlmin 2.0 s a g J l.O0.5 1.5-

O.OK
0 0.02
Punch Figure 5 Load-displacement test specimen during cyclic

0.04

0.06

0.08
(mm)

0.1
punch

Punch Figure 4

Displacement

(mm)

Displacement

Repeatability of the small punch test load-displacement curve for UHMWPE. The four repeat tests were conducted at a rate of 0.5 mm min- on the same day.

behaviour of a small preconditioning.

Relatively little variation in load-displacement behaviour was observed during repeated tests of the specimens at the same punch displacement rate (Figure 4). Mean test results (&one standard deviation) for three testing rates are provided in Table I. Increasing the punch rate significantly increased the peak and ultimate loads, and decreased the ultimate displacement during the tests (P < 0.05). However, the effects of increasing the loading rate, while statistically significant, were not substantial. A fivefold increase in the loading rate (from 0.5 to ~.5mmmin-), increased the peak load by ll%, increased the ultimate load by 9%, and decreased the ultimate displacement by 6%. The work to failure for the small punch specimens was not significantly affected by increasing the loading rate (P > 0.05). Approximately 0.04mm of creep deformation was observed during the cyclic preconditioning, although the slope of the load-displacement curve remained relatively unchanged (Figure 5). After 10 loading cycles, the average initial slope of the load-displacement curve was 61.2 f 4.1 Nmm-. The creep induced by cyclic preconditioning was approximately 1% of the total deformation to failure. Cyclic preconditioning did not significantly affect the peak load, ultimate displacement, ultimate load, or the work to failure (P > 0.05).

Bulk Testing Results (E=851+33 MPa)

tn

200

400
Elastic

600
Modulus

800
(MPa)

1000

1200

Figure 6
initial small

Elastic modulus versus analytically predicted stiffness (up to displacements of 0.064mm) in the punch load-displacement curve.

where the initial stiffness, k, is in Nmm- when the elastic modulus, E, is in MPa. Equation 2 was inverted to provide elastic modulus as a function of initial stiffness: E = 13.5 k
(3)

Analytical
The initial linear slope of the simulated load-displacement curve up to 0.064mm was linearly correlated with the elastic modulus of the material (r2 > 0.99, Figure 6): k = 0.0742 E (2)

Based on Equation 3 and the experimentally determined initial stiffness of the small punch load displacement curve (i.e. 61.2 f 4.1 Nmm-l), the elastic modulus for the four tested CUR4120 specimens was predicted to be 826 f 56MPa. The difference between the elastic modulus predicted by Equation 3 for the small punch test and the elastic modulus measured during previous bulk mechanical tests was not statistically significant (P > 0.05).

Table 1

Effect of testing

rate on small punch test results of Peak load (NJ 60.9 f 1 .O 67.4 f 3.0 69.0 f 1.5 Ultimate (mm) disp. Ultimate (NJ load Work to failure

Rate (mm min-) 0.5 2.5 5.0

Number tests 5 4 4

WJ)

4.63 f 0.06 4.35 f 0.14 4.30 & 0.18

51.9 f 2.8 56.3 * 2.1 55.6 f 2.7

207 f 8 217 + 14 217 i 10

Biomaterials

1997, Vol. 18 No. 24

Small punch testing of UHMWPE: S. M. Kurtz et a/.

1663

DISCUSSION The small punch test is a reproducible miniature specimen testing technique which can be used to characterize the work to failure and ductility for UHMWPE used in total joint replacements. The results of this study demonstrate that the loaddisplacement curve is insensitive to cyclic preconditioning of the test sample and only mildly sensitive to the loading rate. In addition, the initial slope of the small punch load-displacement curve is measurable and ma:y be used to predict the elastic modulus of UHMWPE in conjunction with the inverse finite element method. The mild sensitivity of the small punch test results to loading rate was, consistent with previous bulk mechanical testing results. Stojek and Li15 recently investigated the effect of strain rate on the bulk tensile properties of UHMWPE. Researchers observed no statistically significant changes in elastic modulus and yield stress when the strain rate was increased by an order of magnitude. Although the mechanical response of polymers is well known to be strain rate it appears that UHMWPE does not dependent, exhibit marked strain rate sensitivity at room temperature. Small punch test specimens, despite their miniature size, deformed consistently and homogeneously up to the point of catastrophic failure. The 0.5-mm-thick small punch specimen was nearly on the same length scale as the consolildated UHMWPE resin particles, which typically range in size from 50 to 250mm. However, localized deformation or necking of the specimen was not observed in video tapes or scanning electron micrographs, suggesting that the miniature specimen deformed homogeneously as a continuum during testing. The small punch test is a promising tool by which to directly evaluate UHMWPE mechanical properties in as-sterilized and as-retrieved components. Further research is needed to demonstrate the reliability and reproducibility of the small punch test following sterilization and accelerated ageing. The long-term goal of this research is to develop a miniature specimen testing technique capable of characterizing mechanical properties at the ar-ticulating surface of UHMWPE components for direc:t comparison with in vitro and in vivo wear performance.

REFERENCES
1.

Foulds, J. R., Woytowitz, P. J., Parnell, T. K. and Jewett, C. W., Fracture toughness by small punch testing.
J. Testing Eval., 1995, 23,3-10.

2.

Mao, X., Saito, M. and Takahashi, H., Small punch test to predict ductile fracture toughness JIc and brittle fracture toughness KIc. Scripta Metal Mater., 1987, 25,
2481-2485.

3.

Mao, X., Shoji, T. and Takahashi, H., Characterization of fracture behaviour in small punch test by combined
recrystallization-etch method and rigid plastic analysis. J. Testing Eva]., 1987, 15,30-37. Sutula, L.C., Collier, J.P. Saum, K.A. et al., Impact of

4.

gamma sterilization on clinical performance of polyethylene in the hip. Clin. Orthop., 1995,319,28+0.
5. Collier, J. P., Sperling, D. K., Currier, J. H. et al., Impact of gamma sterilization on clinical performance of polyethylene in the knee. J. Arthroplasty 1996,11,377-89. Kurtz, S. M., Rimnac, C. M. and Bartel, D. L., Degradation rate of ultra-high molecular weight polyethylene. J. Orthop. Res., 1997, 15,57-61. Bostrom, M.P., Bennett, A. P., Rimnac, C.M. and Wright, T. M., The natural history of ultra-high molecular weight polyethylene. Clin. Orthop., 1994,309,20-28. Rimnac, C. M., Klein, R. W., Betts, F. and Wright, T. M., Post-irradiation aging of ultra-high molecular weight polyethylene. J. Bone Joint Surg., 1994, 76-A, 1052-1056. Kurtz, S. M., Rimnac, C. M., Li, S. and Bartel, D. L., A bilinear material model for ultra-high molecular weight polyethylene total joint replacements. Trans. 40th ORS, 1994, 19,p. 289. Kurtz, S. M., Rimnac, C. M., Santner, T. J. and Bartel, D. L., An exponential model for the tensile true stressbehavior of as-irradiated and oxidatively strain degraded ultra-high molecular weight polyethylene. J. Orthop. Res., 1996,X4, 755-761. Kurtz, S. M., Jewett, C. W., Moalli, J. E., Vogt, R. M. and Edidin, A.A., The true ultimate stresses and fracture morphology of ultra-high molecular weight polyethylene upon tensile failure. Trans. ASME Bioengineering (Summer), 1997, pp. 57-58. Engelmann, B., NKE2D. a Nonlinear, Implicit, TwoDimensional Finite Element Code for Solid Mechanics. Lawrence Livermore National Laboratory Report UCRL-MA-105413, Livermore, CA 1991. McKellop, H. A. and Clark, I. C., Evolution and evaluation of materials-screening machines and joint simulators in predicting in viva wear phenomena. In Fundamental Behavior of Orthopedic Biomaterials, ed. P. Ducheyne and G. W. Hastings, 1984, pp. 51-85. Kurtz, S. M., Jewett, C. W., Moalli, J.E. and Edidin, A. A., An elastic-plastic material model for the true stress-strain behavior of ultra-high molecular weight polyethylene in tension and compression. Trans. ASME Bioengineering (Winter), 1997 36, 311-312. Stojek, M.D. and Li, S., The effect of strain rate on the tensile properties of ultra-high molecular weight polyethylene. Trans. 21st Sot. Biomater., 1995, p. 385. Lykins, M.D. and Evans, M. A., A comparison of extruded and molded UHMWPE. Trans. 2lst Sot. Biomater, 1995, p. 112.

6.

7.

8.

9.

10.

11.

12.

13.

14.

ACKNOWLEDGEMENTS Special thanks to J. Moalli, R. Windmiller, D. Crane, L. Pruitt, and S. Valenty for their contributions. Supported by a Research Grant from Osteonics Corp.

15.

16.

Biomaterials

1997, Vol. 18 No. 24