The Mystery Of Physical Aging, Part 3: Accelerated Aging Tests

This is part two of a three-part series discussing the physical aging of amorphous polymer products. Click the links below for more: 

• Part One: Knowing the Difference
• Part Two: Slow Down Aging

---

In the previous parts of this series, we established that physical aging occurs in amorphous polymers at a rate that is dependent upon the proximity of the application temperature to the glass-transition temperature (Tg) of the polymer. The closer we are to the Tg, the more rapidly physical aging occurs. We also know that at a given application condition, physical aging takes place more rapidly in lower-molecular-weight grades of a particular polymer. Finally, we know that physical aging has a measurable effect on mechanical properties. Specifically, it increases the strength and stiffness of material while at the same time decreasing its ductility.

This has some important implications in interpreting the results of accelerated aging tests. These tests have been used for many years to make predictions about long-term performance at particular temperatures based on short-term performance at higher temperatures. The UL relative thermal index (RTI) is an example of such a strategy. Samples are aged at several temperatures for extended periods of time and the physical properties are monitored. At some point, the properties begin to decline, and when they have reached a certain percentage of their original values the material is considered to have failed.

Plotting the log of time to failure as a function of temperature results in a nearly straight line that is then extrapolated to some longer time frame and lower temperature. The slope of this line is governed by an Arrhenius relationship and each successive change in temperature will alter the reaction rate of the degradation process. The typical rule is that for every increase of 10°C (18°F) the reaction rate doubles and the material will then exhibit half the useful life. This is an exponential function, so if the test temperature is changed by 30°C (54°F) the reaction rate will change by a factor of 23 or 8.

This same methodology has been codified into documents that govern protocols for medical devices where extended shelf life at room temperature is a requirement. Using a slightly adjusted acceleration factor of 2.3, aging is performed at a temperature of 55°C (131°F) to accelerate the process that supposedly will occur at 25°C (77°F). By cubing 2.3 we get 12.167, which leads to the tidy conclusion that a 30-day month at 55°C produces the same changes in material properties as one year at 25°C. Implicit within this protocol is the assumption that the only process that is at work on the material is chemical aging, also known as oxidation.

But what if another process acting on the polymer gets into the mix? What if an amorphous polymer with a low Tg, such as PET, is exposed to a temperature of 55°C? We already know that it will undergo physical aging as well as chemical aging. And we know that some of the property changes produced by these two aging mechanisms look the same—specifically, that impact resistance will decline. Loss of toughness in plastic components used in medical devices is obviously a big concern. The question then becomes this: How does physical aging accelerate as a function of rising temperature? 

As it turns out, we know something about this as it relates to PET in particular. Some years ago a major manufacturer of PET studied the effects of temperature on the rate of physical aging using a technique known as differential scanning calorimetry (DSC). At the same time, this company monitored the tensile strength and the impact resistance of the material.

The studies found that for each increase of 10°C the rate of physical aging increased by a factor of 9.8. We repeated these experiments sometime later and came up with a factor of 9.9. In other words, while raising the temperature 30°C increases the rate of oxidation by a factor of 12, this same temperature change increases the rate of physical aging by a factor of nearly 1,000 (9.83 = 941 and 9.93 = 970).

Therefore, the effects of physical aging on the properties of PET dwarf the effects of oxidation. One year at 25°C equals fewer than 10 hours at 55°C when physical aging is the focus of the evaluation. Inversely, 60 days at 55°C produces the same amount of physical aging as would occur at 25°C in 155 years if we assume that there is not some limit to the free-volume reduction that characterizes the mechanism.

The problem is that the documents that govern accelerated testing protocols do not consider physical aging in their protocol. All the property changes that are documented are attributed to oxidation. The procedures attempt to get around these conflicting mechanisms by advising that aging temperatures should be kept below 60°C (140°F) and that this will alleviate any concerns. But as the PET studies show, this is not always the case.

Some of this error is mitigated by the fact that the property that is supposed to be the focus of the aging evaluation is tensile strength. Physical aging will not produce a decline in tensile strength while oxidation will. But both processes have the potential to reduce toughness and physical aging can do so much faster. If a product becomes brittle, it is difficult to convince the end user that everything is fine because the tensile strength has not declined. Loss of impact performance is a very tactile criterion that can produce device failures along with ancillary safety hazards from sharp edges, flying debris and leaking fluids.

The rate of change in properties is not necessarily equal to the rate of physical aging. It has been shown that high-molecular-weight systems will maintain their performance for a significantly longer period of time under the same exposure conditions. But if a manufacturer decides to pursue a lower-MW polymer as a new alternative to an existing high-MW grade, the change in the rate of property decline can be breathtaking.

This happened in a case where an injection molded product was being manufactured in a PET with an intrinsic viscosity (IV) of 0.74 dl/g. This is a fairly high-MW material and can easily fulfill the requirements of thin-walled water bottles made by reheat stretch-blow molding. A new material was offered that had an IV of 0.56. This equated to a reduction in the MW of the polymer of about 35 percent. As molded, the two materials exhibited comparable performance. However, after a relatively short exposure time at 55°C, the parts molded in the lower-MW grade began to exhibit brittle fractures that had never been observed in the high-MW material.

Because the accelerated testing protocol focuses on chemical aging, the assumption was made initially that the loss in performance was related to an inadequate level of stabilization against oxidation, when in fact it was a fundamental response of the lower-MW polymer due to a more rapid rate of physical aging. Increasing the level of antioxidant in the material would not only be costly but it would have no effect on the problem.

This is an excellent example of how important an understanding of fundamentals is to timely problem-solving. Despite the lower MW of the replacement material, tests showed no evidence of oxidative degradation during the aging studies. The changes in physical properties were strictly due to a faster rate of physical aging that would never have occurred during actual storage and use.

The result was that a material that may have been suitable for the application was removed from consideration because of a perceived problem created by an inappropriate accelerated testing routine. These types of problems are common in our industry and until we gain a better grasp of the science of accelerated testing, we will continue to experience difficulties in selecting appropriate materials for our applications.