A. L. McKELVEY and R. O. RITCHIE
Department of Materials Science and Mineral Engineering, University of California, Berkeley, CA 94720 U.S.A.
 

ABSTRACT

This paper describes a study of fatigue-crack propagation behavior in the superelastic alloy Nitinol. This work is motivated by biomedical applications, and the current interest to improve the design and performance of medical stents for implantation in the human body. Specifically, the objective of this work is to study the effect of environment on cyclic crack-growth resistance in a ~50Ni-50Ti (at. %) alloy, and to provide the necessary data for the safe life predication of Nitinol endovascular stents. The material selected for this study has been heat treated such that it is superelastic at human body temperature. Characterization of fatigue-crack growth rates has been performed at 37°C on disc-shaped compact-tension samples, in environments of air, aerated deionized water, and aerated Hank’s solution. Results indicate that, in fact, Nitinol has the lowest fatigue-crack growth resistance of metallic alloys currently used for implant-applications.

INTRODUCTION

The aim of this paper is to describe a study on fatigue-crack propagation behavior in the superelastic material Nitinol. The rationale for this work is the increasing interest in the biomedical industry to improve the design and performance of medical stents for implantation in the human body. The titanium-based alloy, Nitinol, is an attractive replacement for the currently used material, stainless steel, due to its corrosion resistance, interesting nonlinear mechanical behavior, and thermoelasticity.

Despite this interest in implementing Nitinol stents, little is known about their mechanical performance. Indeed, there is considerable industrial activity to develop stents and other implant devices using Nitinol, yet this endeavor is currently limited by a lack of understanding of the structural properties of these alloys; in particular, this makes such vital tasks as stress analysis and life prediction extremely uncertain. For example, although the simple tension and compression properties of Nitinol are reasonably well characterized, there is very little understanding or engineering data on its fracture and cyclic fatigue behavior. This information is of extreme importance as failure of a stent in vivo could result in loss of life. Moreover, due to the unusual nonlinear-elastic constitutive behavior of Nitinol, the deformation and fracture behavior under multiaxial or shear loading is essentially unknown. Efforts to predict the deformation and stresses in stents have been made with various constitutive models using conventional finite-element analysis; however, the accuracy of these models has yet to be tested. Accordingly, there is a critical need for studies involving materials science to understand such vital topics as stress analysis, constitutive behavior, fracture, fatigue and life prediction for medical implant devices involving superelastic materials.

BACKGROUND

The Superelastic Effect in Nitinol

Nitinol is a titanium-based thermoelastic material with a composition of approximately 50 atomic % nickel which originally gained fame during the 1960’s for its shape-memory behavior (Fig. 1) [1,2]. This effect is a result of an athermal martensitic transformation which occurs over a particular temperature range by a uniform lattice shear without a change in the material’s composition. Although the shape-memory effect is interesting, it has little to do with stent design. Subsequent discussion will focus on the more relevant stress-induced superelastic transformation.
 
 

Fig. 1:  Schematic illustration of the shape-memory effect.  An austenite sample is cooled through the martensitic transformation.  At this point the sample is deformed to within ~ 10% strain.  After unloading, the sample is heated through the martensite to austenite transformation.  The martensite laths revert to austenite along the original paths, such that the deformation previously applied disappears as the temperature is increased.  At Af the strain is completely recovered, and the material has the same original shape before the deformation portion of the cycle.  This is a result of the material’s thermoelasticity, low symmetry martensite crystal structure, and also due to the role of twinning in the martensite phase.

Nitinol alloys can exhibit superelasticity at temperatures slightly above the austenite finish temperature, A_f. The superelastic effect is attributed to a reversible stress-induced martensitic transformation (Fig. 2); the stress-strain hysteresis is characterized by three distinct parts. The first elastic deformation shows a great stress increase over a small strain range, De₁.  Following this initial deformation, the curve plateaus with little change in stress for a much larger strain range, De₂.  On this plateau, martensite laths nucleate and grow with the preferred martensite variant (that which is most compliant with respect to the tensile axis).¹  After the martensitic

Fig. 2:  Schematic illustration of the superelastic effect.

Nitinol for Stents

Superelastic stents offer an improved alternative over current options for constricted coronary and carotid arteries. Balloon angioplasty is a technique often used in place of by-pass surgery. The balloon is inserted into the vascular system remotely (e.g., via the femoral artery) and inflated at the blocked arterial site. In order to compress the plaque and open the constriction, it is usually required to inflate the balloon to ~5 atmospheres of pressure. This can induce damage into the vessel wall and cause the artery to be more compliant and susceptible to failure [5]. Stenting is similar to balloon angioplasty, as the surgery is non-invasive, however, it has the additional advantage that it provides a rigid permanent support in the artery to reduce the risk of repeated operations. The stent is simply a hollow tube or helical wire which is designed to support the vessel walls and permit blood circulation through the blockage. Since the stent is permanently left in the body in a fixed position, the risk of arterial collapse is avoided. Currently stents are manufactured from stainless steel; however, titanium and tantalum stents have been proposed [6] and are being developed. Application of these devices requires expanding the stent beyond the plastic limit of the alloy, such that it permanently deforms and embeds in the arterial wall. This plastic deformation can greatly reduce the fatigue life of the stent, and therefore make it more susceptible to failure in vivo, which could result in loss of life.

A dramatic advantage Nitinol has over other alloys is its enhanced recoverable elastic strain. While stainless steel has approximately 0.5% available elastic strain, Nitinol has ~8% due to its superelastic transformation (Fig. 2). Nitinol stents would be inserted into the body, similar to stainless steel stents, except they would be self-expanding as a result of the superelastic effect. Because they need not be embedded into the arterial wall by permanent deformation, it is expected that Nitinol stents would have longer fatigue lives, and the risk of failure would be greatly reduced.
 
 

Fig. 3:  Schematic illustration of the advantage of biased stiffness in Nitinol [7].  The relaxed state of the stent is open.  As the stent is compressed into a guide catheter, the force increases.  When the stent is deployed at the location of the constriction, the force decreases along the unloading plateau.  The blood vessel is supported by constant gentle pressure, however the resistance to future constriction is enhanced by the distinct loading curve with greater stiffness.

Another advantage Nitinol has in stent applications is its biased stiffness [7] (Fig. 3). When open, the stent is in its relaxed state by design. As the stent is compressed into a guide catheter, the device is loaded up the superelastic plateau. After the stent is deployed, it opens and the stress is reduced to a value along the unloading plateau. While a gentle pressure is maintained to keep the artery open, any contraction, or blood vessel diameter re-duction, would result in a higher resistance to loading, and therefore giving rise to biased stiffness. This is a result of the different loading and unloading paths in the monotonic constitutive behavior.

MOTIVATION FOR FATIGUE CHARACTERIZATION

Although Nitinol stents offer many advantages compared to current designs, it is critical that these components be designed to avoid mechanical failure. It is widely recognized that cracks always exist in materials, and that components must be tolerant of their presence. Fracture mechanics provides a basis for design in the presence of such cracks using damage-tolerant concepts. This approach is particularly relevant for stents, which are often fabricated by laser machining. This process invariably leaves a distribution of small (10 – 50 mm sized) cracks in the surface. However, despite the obvious need for such approaches from an engineering perspective, damage-tolerant design methodologies have rarely been implemented in the biomedical industry thus far; one example which is known, is that of pyrolitic-carbon mechanical heart valves where critical crack sizes, fatigue life and fracture toughness data have been incorporated into the valve design and fabrication [8]. Damage-tolerant design is actively used in other industries (e.g., aerospace technology) to estimate failure load/critical crack size combinations which are essential for the prediction of a component’s mechanical life.

Although the thermodynamics and phase transformations of the shape-memory and superelastic effects have been widely studied, currently, there is a lack of information and understanding of the micromechanisms that contribute to the fracture resistance of Nitinol. At present, there are only very limited data in the literature which describe crack propagation in Nitinol alloys under monotonic or cyclic loading [9-11]; furthermore, none of the data in these studies characterizes the fatigue behavior of superelastic Nitinol. In the case of stents, cyclic loads would arise from the difference in systolic and diastolic blood pressures and the stress from the contraction of the heart muscle. Indeed, it is likely that the safe life of such stents will be controlled by the time that it takes incipient cracks to propagate to failure. Thus, it is critical that the crack-propagation rates in superelastic NiTi are known in order to determine expected device lifetimes and provide a rational design against failure.

The superelastic transformation in Nitinol is a potential source of resistance to fracture and fatigue. Mechanically induced phase transformations have already been exploited as a method for increasing the fracture toughness in several materials [12]. Stress-induced martensitic transformations are known to occur in austenitic steels and TRiP steels [13,14]. Also, transformation toughening is observed in brittle materials, which greatly enhances their fracture toughness [15-21]. For example, in partially stabilized ZrO₂, metastable coherent tetragonal particles are dispersed throughout a cubic matrix [21]. When a tensile stress is applied, the metastable tetragonal particles undergo a martensitic phase transformation to a monoclinic crystal structure. This phase transformation increases the volume of the particles, which at a crack tip reduces the local stress-intensity, shielding the tip from the applied load.

Studies on the fatigue of austenitic stainless steels [14,22] and partially-stabilized zirconia ceramics [23] have shown that in the presence of an in situ phase transformation, resistance to crack advance can be significantly enhanced. However, in both the latter examples, the transformation involves a significant and positive dilatational component, which, due to the constraint of surrounding elastic (untransformed) material, results in crack extension into a zone of compressed material [24,25]; the transformation in Ti-Ni alloys conversely, involves largely pure shear with only a small, negative volume change [9], and therefore it is unknown if the micromechanisms which contribute to fracture resistance in Nitinol are similar.

MECHANICAL PROPERTIES: CRACK-GROWTH RESISTANCE

In order to develop an understanding of fatigue-crack growth behavior in Nitinol it is necessary to examine crack-growth rates in an inert environment as well as a simulated body fluid, such as Hank’s solution. Accordingly, fatigue-crack propagation rates were measured for a ~55Ni-45Ti (wt. %) alloy at 37°C, 10 Hz, and a load ratio, R, of 0.1, where R is defined as the ratio of the minimum to maximum stress-intensities in the loading cycle (Fig. 4); the specific environments examined were air, aerated deionized water, and aerated Hank’s solution. From these results it is clear that a corrosion fatigue effect in Nitinol is absent at the frequency studied (10 Hz) in Hank’s solution compared to data collected in air. A similar statement can be made by comparing samples tested in aerated deionized water versus air; note, while deionized water is seemingly inert, dissolved oxygen in neutral pH solutions is known to be a corrosive species in fatigue-crack growth of metals [26]. However, despite the presence of oxygen in the aerated solutions and chloride ions in the Hank’s solution, the fatigue-crack growth data are essentially the same for all three environments. The fatigue thresholds, DK_TH, are ~2 MPaÖm, the slopes of the mid-growth regime, or Paris exponents m, are ~3, and the limits of the applied stress-intensity range prior to failure are ~30 MPaÖm in the present study. These data are necessary to predict the lifetime of a Nitinol medical device. In order to calculate the lifetime, a useful empirical fit is applied to the fatigue-crack growth data; specifically, da/dN = CDK^m, where da/dN is the crack growth rate per cycle, C is a constant, DK is the applied stress-intensity range, and m is the Paris exponent. This equation is then integrated, and solved in terms of N, the number of cycles until failure, by substituting in initial and final crack lengths.
 
 

Fig. 4:  Effect of environment on crack-growth rates in superelastic Nitinol.

While the absence of corrosion fatigue in Nitinol, at least at 10 Hz, is encouraging, it must be emphasized that DK_TH was found to be extremely low for these data when compared to other metallic structural materials used for biomedical implants. To demonstrate this point, Fig. 5 is a plot of fatigue-crack growth data from this study compared with other biomedical metallic alloys. Although these data are for moist air environments rather than simulated body fluid, the fatigue threshold was the lowest, and the crack-growth rates were the fastest, in Nitinol. Specifically, DK_TH is ~4 MPaÖm in a Ti-6Al-4V alloy (Boyce and Ritchie, 1998), which is approximately a factor of 2 greater than the NiTi fatigue threshold. Type 316L stainless steel (Pickard et al., 1975) has a DK_TH equal to 6 MPaÖm, and the maximum applied stress-intensity range prior to failure is nearly 70 MPaÖm. The largest difference in DK_TH compared to Nitinol, however, are for the remaining alloys on Fig. 5. The fatigue threshold for a cobalt-chrome alloy, Haynes 25, which is used in many biomedical applications, including cardiac valve prostheses, at R = 0.05, is ~10 MPaÖm (Ritchie and Lubock, 1986)—a factor of ~5 increase compared to DK_TH for NiTi. Furthermore, commercially pure Ti (Li and Thompson, 1993) has a fatigue threshold greater than 10 MPaÖm when tested in air at room temperature.
 
 

Fig. 5:  Comparison of fatigue-crack growth rates for biomedical metallic alloys.  Data for NiTi from present study; Ti-6Al-4V by Boyce and Ritchie, 1998 obtained at 25°C, 50 Hz, R= 0.1, m = 4.3; Haynes 25 by Ritchie and Lubock, 1986 obtained at room temperature, 30 Hz,  m = 3.5; 316L SS by Pickard et al., 1975 obtained at 22°C, 100 Hz, R = 0.3, m = 4.9; and pure Ti by Li and Thompson, 1993 obtained at 25°C, 10 Hz, R = 0.1, m = 9.6.  All tests were conducted in air.

This comparison of fatigue-crack threshold data may be quite relevant, as the architecture of endovascular stents can be very fine. For example, the strut widths are ~250 mm. Therefore, it may be necessary to design to the fatigue threshold and prevent crack propagation, since there exists a small distance (~250 mm) for a crack to grow before failure of a stent strut.

SUMMARY

This paper describes the benefits of exploiting the superelastic effect in Nitinol in biomedical applications, particularly in endovascular stents. The importance of implementing damage-tolerant design concepts has been discussed, where fatigue-crack growth data are necessary for safe life prediction of implanted medical devices. The crack growth rates in Nitinol were measured at 37°C, 10 Hz, and R = 0.1, in air, aerated deionized water, and aerated Hank’s solution. At the frequency studied, 10 Hz, corrosion fatigue did not appear to be operative. However, the fatigue threshold measured for Nitinol was significantly less than DK_TH in other biomedical metallic alloys. Future work will focus on understanding the micro-mechanisms which contribute to fatigue-crack growth resistance in NiTi.

ACKNOWLEDGMENTS

This work was funded by Nitinol Devices and Components, Inc., Fremont, CA. Thanks are due to Drs. T. Duerig and A. Pelton of NDC for supplying the material and for useful discussions.

REFERENCES

1.   W. J. Buehler and R. C. Wiley, U.S. Patent No. 3 174 851 (1965).
2.   C. M. Jackson, H. J. Wagner and R. J. Wasilewski, Report No. NASA-SP 5110, National Aeronautics and Space Administration (1972).
3.   T. Saburi and C. M. Wayman, Acta Metall. 27, 979 (1979).
4.   C. M. Wayman and T. W. Duerig, in Engineering Aspects of Shape Memory Alloys, edited by T. W. Duerig, K. N. Melton, D. Stockel and C. M. Wayman (Butterworth Heinemann, London, 1990), p. 3.
5.   T. A. M. Chuter, C. E. Donayre and R. A. White, Endoluminal Vascular Prostheses, (Little, Brown and Company, Boston, 1995).
6.   J. E. Bramfitt and R. L. Hess, in SMST-94, edited by A. R. Pelton, D. Hodgson and T. Duerig (First Intl. Conference on Shape-Memory and Superelastic Technologies Proc., Monterey, 1994) pp. 435-442.
7.   T. W. Duerig, A. R. Pelton and D. Stockel, Metall 50, 569 (1996).
8.   R. O. Ritchie, J. Heart Valve Disease 5, S9 (1996).
9.   R. H. Dauskardt, T. W. Duerig and R. O. Ritchie, in Shape Memory Materials, edited by K. Otsuka and K. Shimizu (Mater. Res. Soc. Proc. 9, Pittsburgh, PA, 1989) pp. 243-249.
10.   S. Miyazaki, M. Suizu, K. Otsuka and T. Takashima, in Shape Memory Materials, edited by K. Otsuka and K. Shimizu (Mater. Res. Soc. Proc. 9, Pittsburgh, PA, 1989) pp. 263-268.
11.   K. N. Melton and O. Mercier, Acta Metall. 27, 137 (1979).
12.   R. O. Ritchie, Mater. Sci. Eng. A 103, 15 (1988).
13.   G. B. Olson and M. Cohen, Metall. Trans. A 7, 1897 (1976).
14.   E. Hornbogen, Acta Metall. 26, 147 (1978).
15.   R. C. Garvie, R. J. H. Hannink and C. Urbani, Ceramurgia International 6, 19 (1980).
16.   D. L. Porter and A. H. Heuer, J. Am. Ceram. Soc. 60, 183 (1977).
17.   F. F. Lange, J. Mater. Sci. 17, 225 (1982).
18.   A. H. Heuer, N. Claussen, W. M. Kriven and M. Ruhle, J. Am. Ceram. Soc. 65, 642 (1982).
19.   M. V. Swain, R. C. Garvie and R. H. Hannink, J. Am. Ceram. Soc. 66, 358 (1983).
20.   R. J. Hannink, J. Mater. Sci. 13, 2487 (1978).
21.   A. G. Evans and R. M. Cannon, Acta Metall. 34, 761 (1986).
22.   Z. Mei and J. W. Morris Jr., Metall. Trans. A 21, 3137 (1990).
23.   R. H. Dauskardt, D. B. Marshall and R. O. Ritchie, J. Am. Ceram. Soc. 73, 893 (1990).
24.   R. M. McMeeking and A. G. Evans, J. Am. Ceram. Soc. 65, 242 (1982).
25.   B. Budiansky, J. W. Hutchinson and J. C. Lambropoulos, Intl. J. Sol. Struc. 19, 337 (1983).
26.   S. Suresh, Fatigue of Materials, 1st ed. (Cambridge University Press, New York, 1991) p. 356.