The Materials Analyst, Part 50: The case of the disappearing contaminantThe Materials Analyst, Part 50: The case of the disappearing contaminant
January 1, 2002
This series of articles is designed to help molders understand how a few analytical tools can help diagnose a part failure. Michael Sepe is our analyst and author. He is the technical director at Dickten & Masch Mfg., a molder of thermoset and thermoplastic materials in Nashotah, WI. Mike has provided analytical services to material suppliers, molders, and end users for 15-plus years. |
The best time to catch a problem with a material is while the manufacturing process is still going on.
By the time the product reaches the end user one of the critical pieces of the puzzle, the raw material, is usually no longer available; it has all been consumed during the fabrication process. All that remains is the defective product and hopefully some good parts from another run to use as a benchmark. The absence of raw material is especially problematic when there is a strong suspicion that the material is the cause of the failures. Working back from the product can be an exercise in detective work, as this month's topic illustrates.
In this case, the part was a component made of a 33 percent glass-reinforced nylon 6/6. During a secondary assembly process performed by the molder's customer, a significant hoop stress was routinely applied. Normally this stress was not a problem for the part, but quite suddenly parts began to break on the assembly line.
The material was supplied by a compounder. Compounders typically produce functional equivalents to grades marketed by the major material suppliers, but they do not produce the actual polymer used in the formulation. Because of this, and the fact that they are usually low-cost producers, compounders are very often the focus of any quality issue that appears to be related to raw material. In this case, the molder was so convinced of the raw material's culpability that it had temporarily changed to a more expensive material from a major supplier until it could determine a root cause for the failures.
Preliminary Theories
A few early observations could be made without the use of laboratory equipment. First, the failing parts had a much more resin-rich surface appearance than the good parts. This led to a couple of theories. One was that nylon 6 had been blended into the compound. This is a common approach used to enhance the appearance of filled products made from nylon 6/6. The other theory was that less glass fiber had been put into the material used to make the bad parts. This latter idea had a problem since the part weight had not changed. If glass fiber were actually missing, it would have to have been replaced with an alternate filler that would match the specific gravity of the compound.
There was one observation that did not square with either theory. The fracture surfaces of the broken parts looked slightly porous, and when the parts were first broken, they gave off a distinctive odor that dissipated over a short period of time.
Because the prevailing suspicions centered on composition, a good and a bad part were first compared by DSC for polymer composition and by TGA for filler content and type. In many instances, infrared spectroscopy might be used for polymer composition, but in this case the suspected contaminant was another type of nylon. Nylon 6 and nylon 6/6 are virtually indistinguishable by infrared but can be identified by their different melting points.
Figure 1 (above) shows the first heat results from the DSC. The bad part shows a melting point suppression of 9° C, which can be consistent with a blend of nylon 6 and nylon 6/6. On cooling, the material from the good part recrystallized near 230C, which is typical for nylon 6/6. The recrystallization of the material from the bad part did not occur until the material cooled to 190C, which is actually characteristic of many pure nylon 6 materials. On second heat the melting point of the good part remained near 260C, but the melting point for the bad part had fallen to 219C, again resembling a straight nylon 6.
The theory of a mixture of nylons was looking good. Melting point and recrystallization point suppression can occur in mixtures of nylon 6 and 6/6, depending upon the ratios of the materials and the method of blending. And an improved surface finish is typically the result of incorporating nylon 6 into a nylon 6/6. However, this difference in the parts did not explain the brittle behavior. Nylon 6 and nylon 6/6 are usually compatible and are used successfully as blends in many commercial compounds.
Contaminant Detected
The TGA test, which was intended to verify filler content, began to point to a deeper problem. Figure 2 (below) shows a comparison of the weight loss process for samples taken from the good and the brittle part. The filler levels were fine, as the part weight studies suggested. The good part contained 1 percent more filler than the brittle part, but both results were well within the typical tolerance limits for a 33 percent filled material. An examination of the residues under a microscope showed that they were both composed entirely of glass fibers, and the fiber lengths in each sample were comparable.
The problem was at the low-temperature end of the scan. The good part displayed a decomposition process that was normal for nylon, but the brittle part exhibited an early weight loss that accounted for approximately 3.5 percent of the compound.
There is an important point here. Often when filler content is suspected as a problem, the traditional test to verify the filler level is a simple burn-off of the resin in a muffle furnace. A sample is weighed, loaded into the furnace, and heated to a temperature sufficient to decompose everything that is organic. The residue is then weighed and the filler content is taken as a simple ratio of the final mass and the initial mass. If the furnace method had been used here, the test would have shown no problem with the brittle part. However, the TGA allows the analyst to examine the weight loss process. In this case, it proved to be crucial.
Figure 3 (above) shows the weight loss process for the brittle part along with the derivative curve. The peak weight loss rate was achieved at 340C. There are very few commercial polymers that decompose at this temperature in the TGA instrument. Among the most common are EVA copolymers, acetal, and PVC. All of these materials, when mixed with nylon 6/6 and processed at melt temperatures of 525 to 550F, would rapidly degrade. In the process of degrading, they would form byproducts that would readily break down the nylon.
A check of retained molecular weight using melt flow rate did reveal a large level of degradation in the bad part. The melt flow rate of the good part was approximately 25 g/10 min. The melt flow rate of the material from the brittle part was 270 g/10 min. If you have followed this series, you know that this difference represents a massive amount of degradation and is sufficient to provide a root cause for the poor performance of the parts. But what caused the degradation?
An examination of a library of TGA results showed that the early weight loss process in the brittle part was a perfect match for PVC. A quick test was performed to check for the presence of chlorine in the two parts. This test is called a Beilstein test, and it involves heating a sample of material in a flame while the sample is supported by copper. If chlorine is present, the flame burns green. The test does not quantify the amount of chlorine, but it does determine its presence. The Beilstein test showed chlorine in the brittle part but none in the good part.
The presence of chlorine by itself was not proof of the presence of PVC. Chlorine is also found in many flame-retardant systems. However, in a flame-retardant system that uses chlorine there are also other compounds present such as antimony oxide or zinc borate that act as synergists in the FR package. A more quantitative test by X-ray fluorescence confirmed the presence of the chlorine and quantified the level at 2 percent. More importantly, this test failed to detect any antimony, zinc, or anything else that might suggest the presence of a flame-retardant system.
An attempt was made to positively identify PVC by running an infrared spectrum on the good and the bad part. Unfortunately, the spectrum for even a healthy PVC is very weak. The few bands that do appear are masked by stronger absorptions in the nylon spectrum. Figure 4 (above) shows the results of this test. The only feature that emerged as a difference was a band at 1723 cm-1 in the bad part that is consistent with the formation of degradation byproducts in nylon.
Positive Identification
At this point, we had confirmed an additional ingredient in the brittle part that decomposed in a temperature region consistent with PVC. We had also confirmed massive degradation in the brittle part by viscosity measurements, and we had found chlorine as a residue in the bad part. But how could we confirm the root cause as contamination by PVC?
When polymers become mixed and are hard to distinguish within a mixture, often the best approach is a chemical separation. This is usually achieved by employing a solvent that dissolves one of the polymers in the mixture while leaving the other as a solid. The undissolved material can then be filtered, dried, and identified by infrared spectroscopy. Unfortunately, in this case it was apparent that whatever had caused the problem had been altered in the process so that it no longer resembled the initial material. The good sample dissolved completely, leaving only the glass fiber as a solid. In the brittle part, an undissolved residue was collected. However, it was an insoluble black particulate that proved difficult to identify.
The problem with positively identifying the original cause of the nylon degradation was that the contaminant itself had been changed during the process. This is not entirely surprising. Degradation is a chemical process that changes the chemical structure of a polymer. If an external agent is the cause of this degradation, this agent usually participates in the reaction and is itself altered. In this case, if the culprit was PVC at a low level of 3 to 4 percent, it would have rapidly broken down at the melt temperature conditions used to process nylon 6/6.
When PVC decomposes, it gives off hydrogen chloride, and the PVC polymer begins to change into something that would not be recognizable as PVC. The worse the degradation becomes the more complicated the mixture of breakdown byproducts becomes. In the meantime, the hydrogen chloride rapidly attacks the nylon, leading to the variety of symptoms that our brittle parts displayed. The suppressed melting point, the early weight loss by TGA, the additional band in the infrared spectrum, the large decrease in melt viscosity, and even the internal porosity are all consistent with a process that could have occurred under the proposed scenario. But the only detectable remnant was the chlorine. As an element the chlorine may react and recombine, but it cannot be destroyed.
With cost in mind, the most sensible scenario in a case like this is to intentionally recreate the proposed contamination process and then analyze the results to see if they agree with what has been observed in the initial case.
In this case, that exercise was not considered worthwhile because the proposed root cause was considered plausible. The processor did, in fact, run PVC in the same plant where the nylon parts were being made. The level of apparent contamination was low enough to avoid the catastrophic events that would be typical if pure PVC were exposed to barrel temperatures of 550F. All of the things that would be expected to happen to nylon if it were exposed to hydrogen chloride at elevated temperatures had happened. And the part weight had not changed measurably because the specific gravity of rigid PVC is very close to that of 33 percent glass-filled nylon 6/6.
You May Also Like