Medical Plastics and Biomaterials Magazine | MPB Article Index Originally published March 1996 LECON WOO, JOSEPH PALOMO, MICHAEL T. K. LING, EDDIE K. CHAN, AND CRAIG SANDFORD

March 1, 1996

12 Min Read
Shelf-Life Prediction Methods and Applications

Medical Plastics and Biomaterials Magazine | MPB Article Index

Originally published March 1996



One of the most important tasks for anyone involved in product design and manufacturing is to ensure that a product will function as intended when used by the customer. For the medical device industry, in particular, concerns over both patient safety and potential liability demand that manufacturers offer the absolute maximum in assurances that failures will not occur during use. At the same time, cost-containment policies and competitive pressures often dictate that the least-expensive material and manufacturing process be chosen. These two opposing forces--reliability and cost-effectiveness--together place heavy demands on the designer to satisfy both criteria simultaneously.


No matter how much care is taken in areas such as material selection, tool design, and fabrication, the fact that a product meets all functional requirements upon initial testing does not guarantee its ultimate success. Before arriving in the hands of the end-user, a product must often be sterilized (frequently with ionizing radiation), shipped through various distribution channels, and subjected to shelf storage under varying environmental conditions-- all factors that can induce changes in a product and thus affect its performance. In the case of the alteration of a plastic material over time, real-time aging is undoubtedly the best way of accurately determining long-term performance. But real-time testing is impractical and arduous to perform. Given these difficulties, the use of accelerated-aging techniques based on a fundamental understanding of polymer behavior offers the possibility of obtaining the most reliable predictions of performance short of a total reliance on real-time observations and continual evaluation.

This article examines several commonly used acceleration techniques for shelf-life prediction, explores the basis for their validity, and cites several of the authors' own cases in which successful predictions were achieved. The fact that each case involved a different accelerating method serves to highlight the importance of understanding the basic science when designing the accelerating experiments and of maintaining concurrent real-time product-sample archives to substantiate and validate the method selected.


One of the most widely used test methods is that of thermal acceleration. This technique is based on the application of the Arrhenius kinetics rate equation--the so-called photographers' rule, which states that the rate of chemical reactions roughly doubles for every 10°C rise in temperature. Figure 1 depicts the ratio of reaction rate constants based on the Arrhenius relationship. However, before passing judgment on the validity or accuracy of the approximation as a predictive tool for a particular application, one needs to examine a number of general product-performance factors. These include (1) structural/mechanical values such as modulus, impact toughness, yield and ultimate strengths, and ultimate elongation; (2) optical properties such as haze, yellowness index, and gloss; (3) surface characteristics, including critical surface tension, wetting properties, and adhesion; (4) biocompatibility factors, among them hard- and soft-tissue compatibility, hemocompatibility, thrombogenicity, and complement activation; and (5) toxicity, both acute and chronic.

Molecular-Weight Alterations. Among the principal effects of ionizing radiation are polymer chain scission and cross-linking, which significantly reduce or increase molecular weight, respectively. Either of these processes can result in profound effects on polymer mechanical properties.

Low-Molecular-Weight Species Diffusion. Low-molecular-weight species--either inherent in the polymer or created from ionizing radiation during sterilization as a result of miscibility phenomena--tend to migrate with the passage of time toward the air/polymer interface. At the same time, oxygen diffusion toward the interior of the polymer has the effect of depleting antioxidants or combining with long-lived free radicals from radiation to form peroxy radicals, which has the effect of propagating the degradation chain of reactions and thus weakening the polymer over time.

Physical Aging. The process of physical aging, or free-volume relaxation, is a slow but perceptible densification of amorphous polymers toward the thermodynamic equilibrium state. The rate of aging is greatly accelerated at temperatures close to the glass-transition temperature, Tg. For example, significant aging can take place for PETG (Tg * 80°C) at preconditioning or drying temperatures of 50°­60°C. This physical aging process is frequently accompanied by dramatic changes in the polymer's physical properties.


Physical Aging of PETG. Numerous studies have established that amorphous materials age with time--a process that appears to be universal and irreversible.1 The authors have carried out extensive studies on PETG,2 for which the kinetics of the aging process (as indicated by the recoverable enthalphy determined by differential scanning calorimetry) were found to obey the following relationship:

H = Ha [K­exp (­t/*)]

where Ha is the apparent limiting value of the recoverable enthalpy at temperature Ta, and * is the time constant. The half-life time constant--the time required to attain 50% of the apparent recoverable enthalpy--is plotted in Figure 2 against the departure from the glass-transition temperature (Tg-Ta). The plot suggests that the half-life time increases exponentially with Tg-Ta.

Having determined the extent of aging and the kinetics involved, one merely needs to perform physical testing on samples with a known aging history in order to predict shelf life. In Figure 3, the interdependence of ultimate strength and elongation before and after physical aging is presented. Immediately, one notices the total divergence in the predicted properties before and after aging: elongation decreases drastically while strength actually increases. This seemingly contradictory result serves to emphasize the importance of selecting the most relevant material property for simulation. In most instances (such as the example just cited), the effect of elongation and the associated toughness (a measure of the fracture energy) is most important for product performance. Therefore, elongation should be chosen as the material parameter for prediction.

Cellulose Esters. For a class of cellulose esters suitable for extrusion, it was discovered that shelf-life aging under ambient conditions for very long times (up to 5 years) and high doses of radiation both produced very similar reductions in molecular weight.3 (A separate study analyzing the effect of molecular weight on performance has established minimum molecular-weight standards.) For these materials, a high dose of radiation, or any other suitable method of molecular-weight reduction, can be used to simulate the effect of real-time aging and give a reliable prediction of shelf life.

Irradiated Polypropylene. The authors have previously disclosed significant success in predicting shelf life of irradiated polypropylene.3 Extensive real-time studies and the findings of many other investigators4 led to the conclusion that the mechanical properties of polypropylene degrade with time after radiation sterilization. Evidently, oxygen migration and diffusion into the sample, combined with long-lived free radicals from the radiation, serve to propagate a series of chain reactions resulting in material degradation, even though molecular weight per se is not drastically affected. In addition, the postirradiated sample degradation is extremely heterogeneous in its reaction locus: that is, substantially degraded and embrittled samples will regain most of their ductility and toughness upon remolding. These results led the authors to conclude that the degradation reaction most probably was attacking the tie molecules in the amorphous domains surrounding the polypropylene crystals.

To simulate the postirradiation shelf life, very high surface area per unit mass must be created to facilitate the introduction and diffusion of oxygen. To this end, extruded solid filaments about 1.2 mm in diameter were first drawn and oriented to approximately 80% of the ultimate elongation. In nearly all cases, this resulted in a very porous, ultraoriented state. After irradiation with about 4 Mrd in a cobalt 60 source, testing was carried out immediately at ambient temperature. A specially constructed, multistation creep tester reported previously was used in this study; it is capable of running a maximum of six samples at different stresses simultaneously.5 A microcomputer logs any significant strain variations according to preset criteria. Upon completion of the experiment, creep rupture time tf is plotted against imposed stress on a semilog basis. A typical set of data is presented in Figure 4.

A brief summary of the factors contributing to tf would include: (1) stress imposed on the sample, (2) nominal radiation dose, (3) type of polymer (homo- or copolymer), (4) effectiveness of the stabilizer, and (5) type of radiation (dose rate). Clearly, the objective of using stress as the accelerating parameter instead of temperature has the merit of preserving and more accurately duplicating ambient aging reactions. By using a highly oriented sample, oxygen is freely available for degradation. At the molecular level, the following model can be proposed. The drawing process aligns most of the tie molecules along the stress direction. Some reorganization undoubtedly takes place, in the form of stress-induced recrystallizations and unraveling of the original crystal domains. The end result is a network of microcrystals connected together by fibrils aligned in the stress direction.6 Because the stress applied during the creep experiment is lower than that applied during sample preparation, additional crystalline-domain reorganization is not likely. Instead, the stress is borne at any cross section along the fiber by a large number of polymer chains mostly originated from the amorphous domain. A schematic model of the micromolecular stress distribution is shown in Figure 5.

Given the model just described, the creep-rupture process can be envisioned as follows. Free radicals created during the irradiation process combine with oxygen abundantly available in the matrix to form peroxides that cleave the tie molecules in the fibril under stress. Since all fibrils are mechanically connected in parallel, cleavage of any individual fibril would unload the stress on the remaining fibrils. The resulting higher state of average stress would lead to a greater probability of failure, and the process would accelerate until ultimate failure. Changing the chemical composition (through copolymerization with ethylene) would lead to a greater resistance to cleavage on individual tie molecules, and to readily measured improvements in rupture time. The use of stabilizer molecules would have a similar effect.

The data in Figure 4 can be extrapolated toward lower stresses to indicate failure times at these stresses, providing a quantitative accelerating prediction based solely on molecular behaviors. The validity of extrapolation and possibility of gross nonlinearities at low stress regimes can be verified only experimentally. Toward this end, we used existing real-time data to "calibrate" the approximate stress levels that are likely to exist in stored ambient-aged samples.

The failure at zero stress is the induction period for the auto-oxidation with freely available oxygen. This is the condition that existed in the topmost surface layer of the molded sample. Of course, due to oxygen diffusion, the interior of the thick samples suffered far less degradation. Nevertheless, there are situations in which surface embrittlement could lead to bulk failures, such as upon impact. The slope of the failure time (tf) versus stress line represents the reduction in failure time due to incremental increases in stress. This can be viewed either as an activation process for chain rupture--in which the external stress simply reduces the activation energy of the reaction--or as the distribution of stress-bearing chains capable of withstanding the stress for a given period of time. Added stresses would simply shift the distribution toward a shorter time. From the above analysis, one can deduce that the slope should be dependent on the polymer type, morphology, and chemical effectiveness of the stabilizer system, while the zero-stress intercept would be primarily dependent on the total number of tie molecules and the molar concentration of stabilizer molecules.

An examination of the data proves to be very helpful. In Figure 6, data from the same resin source, sample lot, and processing history were used. Following the orientation process, samples were subjected to (1) no radiation (control), (2) 4 Mrd of gamma at a dose rate of about 0.3 Mrd/hr, or (3) 4 Mrd of a 1-MeV electron beam at a dose rate of about 30 Mrd/min. The nearly identical slopes observed for all three samples indicated that the reaction chemistries are very similar. The gamma-irradiated sample had the lowest intercept at zero stress, indicating substantial degradation during and immediately after the irradiation. The electron beam­irradiated sample behaved significantly better than the gamma-processed sample. This is one of the first reported instances of clear-cut evidence that at higher dose rates, with limited oxygen availability, electron-beam irradiation can result in much longer shelf lives compared with gamma. Undoubtedly, free-radical recombination at higher temperatures as well as limited oxygen availability during the high-dose-rate event all played major roles in protecting the material.


Based on actual simulation experiments, it was concluded that relying on temperature alone to accelerate shelf-life aging conditions is unrealistic. Consideration must be given to critical material-performance parameters and, further, to the molecular basis for the origin of these factors. Properly designed experiments, especially with supporting data from parallel real-time monitoring, will lead to the most realistic predictions.


1. Struik LCE, Physical Aging in Amorphous Polymers and Other Materials, New York, Elsevier, 1978.

2. Woo L, and Cheung YW, "Physical Aging Studies in Amorphous Poly(ethylene terephthalate)(PET) Blends,"Thermochimica Acta, 166:77­92, 1990.

3. Sandford C, and Woo L, "Shelf-Life Prediction of Radiation-Sterilized Medical Devices," in Proceedings of the 45th Annual Technical Conference & Exhibition (ANTEC), Brookfield, CT, Society of Plastics Engineers, p 1201, 1987.

4. Williams JL, "Stability of Polypropylene to Gamma Radiation," Polymer Preprints, 31(2): 318, 1990.

5. Woo L, Sandford C, and Walters R, "Recent Advances in Medical Plastics Analysis," in Advances in Biomaterials, Lee SM (ed), Lancaster, PA, Technomic Publishing, p 52, 1987.

6. Samuels RJ, Structured Polymer Properties, New York, Wiley, 1974.

Lecon Woo, PhD, is the Baxter distinguished scientist in the Medical Materials Technical Center at Baxter Healthcare (Round Lake, IL), where he specializes in biomedical polymer development and polymer rheology and processing. Before joining Baxter in 1982, he worked at Du Pont and Arco Chemical on polymer characterization and product development. A fellow member of the Society of Plastics Engineers, Woo holds more than 14 U.S. and international patents and has coauthored more than 70 technical papers and book chapters. Joseph Palomo is the senior principal engineer in the custom sterile division of Baxter's surgical group, specializing in surgical disposables product development. Michael T. K. Ling is senior technical specialist in the Medical Materials Technical Center. His research field includes medical product development, mechanical and physical analysis, and process development. The late Eddie K. Chan was formerly the senior principal engineer in Baxter's corporate Material and Membrane Technical Center. Craig Sandford, now at Viskase Corp., was formerly associated with the Medical Materials Technical Center, where he conducted extensive research on the effect of ionizing radiation on medical materials.

Sign up for the QMED & MD+DI Daily newsletter.

You May Also Like