The fundamental difference between metals and polymers—molecular structure—provides clues as to why the polymer sometimes fails in the first attempted substitution.
Several years ago I worked with a client on an application for a molded fan. The part was specified in a 30% glass-fiber-reinforced polypropylene homopolymer. The tensile strength of the material was given on the data sheet as 12,000 psi (82.75 MPa). The part geometry followed good plastic part design principles and it was gated in a manner that ensured a relatively even pressure distribution in the mold cavity and minimized the likelihood of warpage.
First parts were produced and placed on a spin test to establish the functional behavior of the product. Within 5 hours the tests registered the first failure. This was orders of magnitude below the expected life of the product. As is usually the case in a situation like this, the molding process came under immediate scrutiny. Suggestions regarding melt temperature, mold temperature, injection speed, and packing pressure were offered from every corner.
However, the client had done some substantial finite-element analysis (FEA) work and I wanted to see if the areas where failure was occurring corresponded to those locations where the model indicated that the highest stresses were present. It was surprising to see that the highest predicted stresses were near 8200 psi (56.6 MPa), and I suggested to the client that the material selected for the application was not strong enough to sustain such high stresses for any length of time. This was a cost-sensitive application, so the last thing that the client wanted to hear was that a more expensive raw material might be required to make a part with the desired lifetime. They produced a stress-strain curve for the material that verified the published tensile strength on the data sheet and more or less considered the matter settled.
The design engineers who developed this part were accustomed to working with metals and had always employed the rule of thumb that a maximum working stress of 70% of yield was an acceptable threshold for ensuring proper performance. They were victims of the assumption that the behavior of plastic materials follows the same rules as those for metals. In addition, even had they been aware of the fundamental differences between these two classes of materials, they lacked the quantitative information needed to make good decisions about the long-term behavior of the polypropylene (PP) over the intended lifetime of the product.
Molecular variation
If we look at the fundamental building blocks in a metallic and a polymeric material, we can begin to appreciate the reasons for the differences in behavior. At a molecular level, the basic units that make up a metallic structure are relatively small and uniform in size. Consequently, they readily arrange into a structure that is very regular and predictable. Material scientists identify these structures as crystalline solids. This type of structure is influenced very little by changes in temperature until the materials in this class reach their respective melting points. Metals are also not significantly affected by sustained loading until the stresses reach a point close to the yield strength of the material. In other words, they are not generally susceptible to creep or cold flow.
The smallest unit in a polymeric material is a very large molecule with an extended chain shape. Even the smallest polymer molecule will be about 50 times more massive than the heaviest naturally occurring unit in a metallic structure. These polymer molecules can twist, turn, fold, and entangle in almost unlimited variations. In addition, not all of these molecules are of the same size. The smallest molecule in any random sample of plastic is often a thousand times smaller than the largest, which increases the possibilities for local variation in the structure of a material.
Figure 1. Melting Behavior of Indium Metal and Polypropylene by DSC
This has important implications for the mechanical and thermal properties of the two classes of materials. Figure 1 shows temperature scans for polypropylene and a low-melting elemental metal known as indium. The two materials have comparable melting points. If you check the literature, you will find that the melting point of indium is 156.61°C, give or take a couple of thousandths of a degree. This property is so consistent and so easily measured that the metal is used to perform temperature calibrations on thermal analysis instruments such as differential scanning calorimeters (DSC) and thermogravimetric analyzers (TGA). Polypropylene has an approximate melting point of 165°C. But if we compare the process by which these two materials melt, we can see a very large difference.
Melting can be considered to have begun when the heat flow vs. temperature plot begins to depart from the baseline and it has concluded when the plot returns to baseline. In Figure 1 we can see that the melting process for indium has a width of less than 2 deg C when the sample is heated at a rate of 10 deg C/min. This means that all of the indium crystals melt at essentially the same temperature because they are all of approximately the same size and shape.
However, the polypropylene melting process spans 45 deg C. This is an indication that the crystal sizes and shapes within the polypropylene sample vary significantly. To make matters worse, virtually all of the indium atoms bound into the macrostructure of the metal are part of a crystal. But the melting process for polypropylene may involve less than half of the structure. So while we refer to metals as crystalline solids, in polymeric materials we can only talk about degrees of crystallinity.
Mechanical performance and stresses
Figure 2 attempts to give a conceptual view of what a crystalline structure looks like in a polymer. Areas of the structure organize in fairly well-defined patterns. These represent the crystallized regions. In some polymers these well-organized regions may constitute as much as 90% of the material. However, in others this level can be so low that it is insignificant for all practical purposes and a melting point cannot even be detected.
Figure 2. Schematic of Semicrystalline Structure
That portion of the material that fails to organize into crystals is referred to as amorphous. These amorphous areas achieve a substantial level of molecular mobility at a much lower temperature than the measured melting point. We perceive this mobility as a reduction in the strength and stiffness of the material. In addition, when placed under load, these amorphous regions tend to deform at stresses much lower than any measured upper limit such as the yield point. Metals, by contrast, contain little or no disorganized material. Therefore, when they are placed under load, their behavior is much more predictable.
This difference has important implications for mechanical performance and how that performance is affected by the influences of temperature and time. Figure 3 shows a classical stress-strain curve for aluminum. In the language of metals, point #2 represents the yield strength of the material. Between the origin point on this graph and the yield point, the relationship between stress and strain, quantified as the modulus, is essentially constant. This is behavior that we typically refer to as elastic. As long as we do not reach the elastic limit, we expect the material to perform up to the expectations defined by measurements of strength and modulus.
Figure 3. Stress-Strain Curve for Aluminum
Figure 4 shows a stress-strain curve for an acetal copolymer. On the scale of 1 to 10, with 1 representing no crystal structure and 10 representing the highest level of crystallinity practically attainable in a polymer, acetals are a 9. However, even a material like this shows a significant amount of nonlinear behavior as it proceeds from the origin up to the yield point. Yield may not occur until the material reaches a strain of almost 15%, but even a visual assessment of this curve shows that the linear relationship between stress and strain has broken down before the curve has achieved 1% strain.
Figure 4. Tensile Stress-Strain Behavior for Acetal Copolymer (23°C)
An expanded view of the curve provided in Figure 5 shows that the proportional limit, the last point on the stress-strain curve where the relationship is linear, occurs at less than 0.5% strain. Stresses that fall between the proportional limit and the yield point will have a greater and ever-changing effect on the resulting strain depending on where we are on the curve and for how long the stress is applied. We are not picking on acetals here; all plastics follow this same general behavior to one degree or another.
Figure 5. Early Portion of the Acetal Stress-Strain Curve
Effects of temperature
Figure 6. Effects of Temperature Changes on the Stress-Strain Behavior of a Glass-Filled PP
It gets better. The operating temperature of the fan was 60°C. The stress-strain curve for a metal like aluminum is relatively unaffected by changes in temperature, at least within reason. Figure 6 shows stress-strain curves for a 15% glass-filled polypropylene at room temperature and at 43°C and 55°C. Over the relatively small temperature range between 22°C and 55°C, the yield stress and modulus of the material have declined by approximately 35%. We can reduce the effects of elevated temperature on the mechanical properties of the polymer by employing higher levels of reinforcement. The fan was produced from a 30% glass-fiber-reinforced material that only sustained a 25% reduction in short-term mechanical performance across a comparable temperature range.
Now let's return to our fan designers. It is apparent from the data in Figure 6 that the mechanical properties of polypropylene, even when reinforced with glass fiber, exhibit significant sensitivity to the effects of temperature. Imagine operating at what is believed to be 70% of the yield stress for such a material, only to find out subsequently that the increase in temperature brought the stresses modeled in the FEA to almost 95% of its short-term yield point.
The engineers did not understand the need for a stress-strain curve generated at the operating temperature of the device. Even if they had known of the importance of this data point, they would have likely discovered that the curve they needed did not exist. And this is before we have begun to consider the effects of extending the application of these stresses to hundreds or thousands of hours.
A treatment of the effects of time will have to wait for Part 3. However, in advance of that installment, consider this: Time-dependent effects such as fatigue, creep, and stress relaxation impose additional demands on plastic materials that typically do not require consideration when working with metals. The bad news is that in most cases we lack good information about this time-dependent behavior in plastic materials. The good news is that we know how to measure and evaluate these influences, if only we will take the time to do it. Next month we will demonstrate these capabilities quantitatively and show how failure to implement these techniques frequently leads to products that are either over- or underengineered.
About
Michael Sepe Michael Sepe has worked in the plastics industry since 1975 in a variety of roles involving both manufacturing and research and development. He is an independent consultant based in Arizona with clients throughout North America. He assists clients with material selection, designing for manufacturability, process optimization, troubleshooting, and failure analysis. Learn
more about Michael Sepe.
