It was established in Part 1 of this discussion that physical aging involves a change in the free volume within the amorphous regions of a polymer matrix. Specifically, it involves a volumetric contraction of the matrix that occurs over time. This should be observable as a change in dimensions as well as properties.
The dimensional changes are usually much more subtle than the property changes. The amount of dimensional change will depend upon several factors. These include the size of the part, the molecular weight of the polymer, and the molding conditions that were used to produce the parts. Detection of these dimensional changes will depend upon the precision of the measurements. We will examine each of these factors here.
Dimensional changes are typically measured as a percentage. In unfilled amorphous polymers, the shrinkage exhibited by the molded part compared with the tool-steel dimensions is typically 0.4-0.8%. The exact value will depend upon the material, the part geometry, and the processing conditions. Anyone who has molded parts to a demanding tolerance knows that the ability to change a molded dimension depends to a significant degree on the size of the part. If a change in processing conditions can alter the shrinkage by 0.1%, then in a part with a length or diameter of 0.250 in., the change will only be one-quarter of a thousandth, a difference that may go undetected if measurements are being made with a tool that has a resolution of 0.0005-0.001 in.
However, if a part is 50-in. long, it may be possible to move this part length by 0.050 in. The dimensional changes that occur due to physical aging are far smaller and occur more gradually than the shrinkage that occurs during molding. Therefore, only large parts or parts that are measured to a very high degree of precision will exhibit detectable changes due to this mechanism.
Even if such changes are noted, it is unlikely that the reason for their occurrence will be understood or properly interpreted. A couple of years ago I worked with a company that made molds out of PVC. The two halves of the mold were themselves injection molded and, when assembled, contained features into which casting resins were poured to form the final product. The dimensions of this cast product were very critical, consequently the control over the dimensions of the PVC parts was critical. Once the process reached equilibrium and parts were being produced that met print specifications, the written procedure for using the parts to create the cast pieces dictated that use could begin 24 hr after molding. These parts could continue to be used until they were four months old, at which point they had to be discarded.
This seemed like an interesting time window, and when I asked about the origin of this guideline I was told that it had been observed over the long history of this part that after four months some of the parts had become undersized. The measurements made on these parts are in terms of microns, not thousandths of an inch, therefore the dimensional changes were more evident. They occurred as rapidly as they did because the glass-transition temperature of rigid PVC is relatively low at 78 C (172 F).
As this example suggests, once the mechanism behind the dimensional change is understood, steps can be entertained to slow down or stop the physical aging process by storing the parts at lower temperatures, thereby extending the useful life of the product. Stopping the physical aging process in PVC requires a temperature below -50 C (-58 F), which is probably not practical. But dropping the temperature at which the parts are stored by just 10° C can slow the physical aging process by almost a factor of 10, making the parts functional for over three years instead of four months.
The rate at which physical aging takes place is governed in part by the molecular weight of the polymer. The reduction in free volume associated with physical aging requires some degree of mobility at a molecular level. Lower-molecular-weight polymers consist of shorter chains that exhibit less entanglement and therefore a greater degree of mobility than higher-molecular-weight systems.
This has been observed in amorphous PET polyester, where the impact performance of a grade of material with an intrinsic viscosity (IV) of 0.53 dl/g displays a reduction in impact resistance at a rate more than three times faster than a grade with an IV of 0.67. A lower intrinsic viscosity is associated with a lower average molecular weight. We will come back to this aspect of physical aging next month in Part 3, as it applies to an accelerated-aging qualification process.
THE ROLE OF PROCESSING
Finally, the molding conditions have an influence on how physical aging affects the long-term behavior of a material. No commercial process designed to produce parts at a competitive price can achieve a structure that is free of internal stress and achieves the perfect equilibrium state. However, some process conditions will produce parts that are closer to this ideal than others.
Slow cooling rates provide the time needed to relieve the stresses that are created by the fabrication process. Rapid cooling does not provide for this relaxation process, and the polymer chains will be trapped in a configuration that is farther from the ideal equilibrium state that the material will try to achieve after the part is molded. Cooling rate is controlled primarily by the temperature of the mold, and the relationship between mold temperature and internal stress is quite well established. All things being equal, higher mold temperatures produce parts with a lower level of internal stress, which can be verified by performing tests on molded parts that measure this stress. In situations where the process alone cannot achieve the desired level of molded-in stress, annealing may be performed.
Unfortunately, processors tend to run their tooling at temperatures that are lower than optimal for minimizing stress and the subsequent effects of physical aging because they believe that a lower mold temperature will yield a shorter cycle time. However, the faster cooling rates afforded by lower mold temperatures do not reduce cycle time as much as is thought, and the loss in performance does not justify the increased production rate. In addition, parts produced with a lower mold temperature often display surface defects such as more pronounced flow lines, gate blush, and inconsistent gloss.
The superior quality of parts produced in a mold set at a higher temperature has been proven repeatedly, but it is still a proposition that is greeted with skepticism by many in the processing community because the difference in quality is often not apparent unless long-term tests are performed.
Three years ago I was involved in sampling a mold producing thick-walled polycarbonate parts. The initial run had been conducted at a mold temperature of 60 F (15 C). The parts had evident flow lines that reduced the natural transparency of the material. This led to a second sampling where the mold temperature was increased to 220 F (105 C). To counter the effect of the hotter mold the melt temperature was reduced from 610 F (321 C) to 500 F (260 C). The cycle time did not change and the part appearance improved dramatically.
But the real benefit to material performance did not become evident until two years after the parts were molded. At that point the parts molded at the low mold temperature began to form cracks at the corners in the absence of any applied load. These cracks continued to grow larger over time, while the parts produced in the hotter mold continued to function and showed no evidence of cracking.
In Part 3 we will discuss the influence of physical aging on accelerated test protocols and show why it is important to consider this mechanism in interpreting the results of these tests.