Physical Testing as a Plastic Failure Analysis Tool

By Michael Sepe, Materials Analyst

Datasheets provide results based on a perfectly molded bar specimen. The real world is always different.

When a molded part fails in the field or in qualification testing, the performance shortfall is almost always a mechanical one. And within this very large subset of possible mechanical failures, the property most likely to disappoint is impact resistance, often referred to as toughness or ductility.

When evaluating the cause of a plastic product failure, analysts have a number of tools at their disposal. There are chemical techniques to check composition, there are a variety of viscosity tests to evaluate the molecular weight of the polymer, and there are direct inspection techniques such as cross sectioning and scanning electron microscopy that allow for an inspection of features that tell us where the failure started, how it traveled, and what may have gotten it started in the first place.

Sometimes attention turns to an actual measurement of the physical properties of the material in the molded part. This involves machining standard test specimens from the molded part and performing tests to gauge the strength, stiffness, and toughness of the material. This avenue can have some value, particularly if known good and known bad product can be supplied for a direct comparison. However, this technique also has a lot of pitfalls because the person requesting the work often has a hidden expectation—that the values derived from specimens cut from the molded part will match those on the data sheet.

There are a host of problems with this reasoning, and the reality is that unless the geometry of the part is as simple as that of a standard tensile bar and allows for the same degree of polymer orientation, there is almost no chance that the properties of a specimen cut from a molded part will match those of a molded test specimen. It should be recognized by one and all that the properties obtained in a standard test specimen reflect a best-case scenario when it comes to measuring properties such as tensile strength, flexural modulus, and notched Izod impact strength.

First, an examination of a mold designed to produce test specimens shows that the gates and runners are generous, to say the least, reducing shear stresses on the material and decreasing the resulting damage that can occur, particularly to fibrous reinforcements.

Second, the gate is placed on one of the bar’s short ends, ensuring that maximum orientation will occur in the direction in which the sample is tested. Even unfilled materials can benefit from this type of preferred axial alignment, but when glass fibers are involved, properties in the direction of flow can be 2-2.5 times greater than in the cross-flow direction.

Finally, the specimens have a uniform nominal wall, and nothing else. There are no ribs to divert or interrupt polymer flow, no thick sections that create localized molded-in stress, and no cored-out areas that create weldlines. So when design engineers surf through property databases and find a 43% glass-fiber-reinforced nylon with a tensile strength of 33,000 psi, it is imperative that they understand that this value will never be realized in the real world of complex molded parts.

A Dance With Fillers and Moisture
Recently, one of our clients attempted to gauge the quality of a molded part using physical property measurements in order to determine the cause of occasional breakage in the field. This was a large part, and while not perfectly flat, it did provide a geometry that allowed for the creation of samples that were a reasonable approximation of the standard tensile, flex, and notched Izod specimens.

The material was a 20% glass-fiber-reinforced nylon 6 and the product is exposed to a considerable amount of incidental abuse during its life. Tensile, flexural, and notched Izod properties were evaluated. Fortunately, the wall thickness of the part—3 mm (.118 inch)—was close to that of a standard ASTM specimen—3.175 mm (.125 inch).

Anyone who has conducted notched Izod tests on samples of varying thickness knows the importance of this parameter. However, the flow path of the material entering the portion of the cavity where the geometry was favorable for sample preparation did not allow for perfect orientation of the glass, and this was the first factor that created a discrepancy between the literature values and the real world. But the most interesting complication came from the base polymer.

As almost everyone knows, nylons are very hygroscopic, particularly the workhorse materials nylon 6 and 6/6. We are not talking about hygroscopic as in polycarbonate or PBT polyester, where equilibrium values are near .1% and saturation values approach .3%. Nylons can and will absorb amounts of moisture that are an order of magnitude higher than other polar polymers. More importantly, when they do, their physical properties at room temperature change significantly.

In an unfilled material, strength will decline by 40-60% as the part goes from the so-called dry-as-molded state (less than .2% moisture by weight) to a conditioned state (approximately 1.5-2.5% moisture by weight, depending on whom you talk to). Modulus may decline even more dramatically, by 60-80%. And as is almost always the case, when load-bearing properties decline, those attributes associated with ductility increase. For better or worse, in our world of plastic properties the standard impact test method is still the notched Izod test, and notched Izod impact strength will more than double as the nylon material gains moisture.

The presence of fillers and reinforcements mitigates these changes because the filler displaces the hygroscopic polymer with a nonhygroscopic, inorganic material like glass fiber. But the changes are still measurable. For this reason, many material suppliers publish two data sheets for nylon compounds to cover the range of properties that result from the big changes in moisture content. So we can observe how the presence of fillers alters the response of the material to these changes by studying these data sheets. Table 1 (p. 26) captures some typical values for nylon 6 with various amounts of filler.

The Real World vs. Data Sheets
The failed part in question had been in the field for nearly a year, more than enough time for the polymer to come to equilibrium with the moisture in the atmosphere. The product is intended for indoor use, where relative humidity values can fluctuate from 30% in the winter to 70% in summer. A Karl Fischer titration showed that the part contained 1.25% moisture by weight. Remember that this compound is only 80% nylon, and when this moisture content is adjusted for the glass content, it calculates out to 1.56% moisture by weight of the available polymer.

How much will this affect the properties of the material? Unfortunately, this particular material supplier did not publish a data sheet for both the dry-as-molded and the conditioned state. So the best we can do is refer to the table and interpolate an expected result for the 20% reinforced product from the effects that we see on the 15% and 30% filled grades.

Table 2 shows the measured properties from the various test specimens compared with the only available data in this grade of material from the literature.

How do we interpret these data to determine if we have a fundamental problem with material performance? A straight comparison based on only the effects of the absorbed moisture would lead us to expect a decrease in tensile strength of about 35% and a reduction in flexural modulus of about 25%.

Both of the actual results are reasonably close to these values, and the 6-8% offsets can be more than accounted for by the differences in orientation between the machined specimens and an ideal molded test bar.

The one apparent anomaly, which might be seized on by someone with an agenda, is the reduction in notched Izod impact strength. A study of data sheets suggests that the notched Izod values should more than double as the result of moisture uptake and the dry-as-molded value of 1.8 should go up to 3.5 or 4.

But Izod impact strength values are also significantly affected by glass fiber orientation. An end-gated notched Izod bar has glass fiber aligned across the notch that is machined into the bar, as shown on p. 27. When the pendulum strikes the specimen, much of the force required to break the specimen goes into breaking through all of these aligned fibers. This tends to give false high readings in glass-filled materials, where alignment follows the direction of flow.

This can be seen if we look at the difference in Izod impact resistance between an unfilled and a glass-filled nylon. A dry-as-molded unfilled nylon 6 has a notched Izod impact strength of about 1 ft-lb/in. Adding 30% glass to the material raises this value to 2.5. The same effect can be observed in comparing short-glass materials to analogous long-glass materials. A 30% short-glass-reinforced nylon has a notched Izod impact strength of 2.5 ft-lb/in, while the same amount of long glass boosts that number to 9 or 10.

This is not a realistic reflection of how glass fibers influence real toughness; it is an artifact caused by the effect of more aligned fibers crossing the notch area. A real-world evaluation of these materials shows that the addition of glass fiber reduces toughness in a material and even the apparent gains reflected in the long-glass materials turn out to be modest when more realistic assessments of impact strength are made. Therefore, the random alignment of fibers in our samples can substantially reduce the result shown in the data sheet.

Straining For the Answer
So how do we get a better handle on real toughness? One useful metric is the amount of strain that a material can withstand before it breaks. During a tensile test, a specimen can be stretched until it breaks into two pieces. The amount of strain required to produce this failure is referred to in the literature as elongation at break or ultimate elongation. This is an important reality check on notched Izod values.

Note that the elongation at break in the samples cut from the molded part is three times greater than the value on the data sheet for a dry-as-molded specimen. If we go back to the data sheets and assess the improvement that we would expect solely from moisture gain, it comes in between a twofold and a threefold improvement. So at the very worst, the improvement in practical toughness is in line with what we would expect, and there is no reason to suspect, based on physical properties alone, that this part is deficient in performance.

The difficulties involved with sorting through all of the effects of part geometry and application environment show that this is not a productive approach to assessing part and material quality unless two classes of product with verifiably different performance are available. And even then care must be taken to ensure that test specimens are prepared from the same location on the part to avoid influences from features like wall thickness changes and weldlines. Typically, if a problem exists that relates to molding or material quality, it will show up in tests for composition and molecular weight, or in a direct examination of the fracture. These tests are far less ambiguous in their interpretation.

July , 2006 - Reprinted with permission from Injection Molding Magazine. Copyright © Canon Communications LLC.