Sponsored By

February 1, 2000

12 Min Read
The Materials Analyst, Part 29:  Defining responsibility for field failure

This series of articles is designed to help molders understand how a few analytical tools can help diagnose a part failure problem. 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 the last 12 years. He can be reached at (414) 369-5555, ext. 572.

We would all like to think that in these times of increased competition, cooperative efforts between suppliers and customers are an important part of the new business climate and a key tool in the process of solving problems rapidly. Unfortunately, all too often the large end user, when confronted with a product performance problem, no longer possesses the in-house expertise or the inclination to assist in the problem-solving process. It often turns to the supplier for the solution.

Frequently the implication of the request for help is that the supplier has in some way caused the problem, and a solution is needed yesterday or else. In such a situation, where guilt is presumed, the only salvation for the supplier may be an objective analysis of the problem. The responsibility must land somewhere, but in order for the solution to be a lasting one, the determination must be based on data and not on opinions.

In a world where the design and material selection process is pushed further and further down the supply chain, it becomes imperative that the general specifications provided by the end user accurately reflect the end-use conditions that the part will see. This particular case involved an automotive part in an under-the-hood application where the environment was oil immersion at elevated temperatures. The general criteria called for the part to have good ductility, moderate strength and stiffness, and an ability to handle oil immersion at a constant temperature of 110C (230F) for the anticipated life of the product, approximately 4000 hours. Given these specifications, and using past experience with successful applications, a heat-stabilized, impact-modified nylon 6/6 with 14 percent glass fiber was chosen for the part.

Testing to Extreme
At this point we come to our first problem. Engineers like to perform what they refer to as reliability tests. Often these involve running the application at very aggressive conditions in order to accelerate the deterioration of possible weak links in the system. The presumption is that if a product can withstand a punishing environment for a certain amount of time, then it is certain to sustain the normal exposure conditions for the anticipated life of the product.

The test bed conditions for this particular product were not specified, but when one part being tested under an accelerated protocol failed in a brittle manner during a routine check, the investigative machinery was set into motion with a vengeance. The molder that had produced the part received a portion of the broken piece back with a stern letter demanding an explanation and a corrective action plan. Also requested was a mountain of technical data concerning maximum and minimum time-temperature performance profiles, information that should have been requested and studied before the application was launched. We in turn received the same package of information from the molder with the assignment of finding out what went wrong.

Visually there was an obvious problem. The part that went into the application was white, but the piece that we received was a very dark brown. This suggested the presence of a significant amount of heat. Nylons are well known for their tendency to change color when they are exposed to elevated temperatures. Even in a resin dryer, nylon pellets that are dried at high temperatures begin to darken as they undergo a chemical reaction known as oxidation. This oxidation process eventually results in a breaking of the polymer chains and a reduction in molecular weight. As we have mentioned repeatedly in previous installments of this column, reduced molecular weight compromises properties, especially ductility and impact resistance.

Figure 1 compares the melt-flow rate of two nylon parts aged at two different temperatures. One part was aged at 80C (176F) while the other part was aged at 120C (248F) for the same period of time, 11 1/2 days.

In that short time, the part aged at 80C retained its as-molded molecular weight, while the part aged at 120C exhibited serious problems early on. By the end of the test, the melt-flow rate of the latter part had tripled, signaling serious degradation. Property tests performed on each of these parts showed a significant decline in impact strength for the degraded part but virtually no change for the part aged at the lower temperature.

Identifying the Chemical State
The selection of the two temperatures in this study was not an accident. The product was a general-purpose nylon with no special stabilizers. Materials of this type can withstand a maximum temperature of only 80C without breaking down. In our client’s application, the added heat stabilizer was designed to extend this protection up to 120C, but above this point the data from the material supplier showed that the rate of property decline would increase rapidly. The appearance and physical condition of the sample led us to believe that the part had seen temperatures well above the prescribed 110C.

Normally, melt-flow tests would be used to confirm the presence of degradation as they were in the study shown in Figure 1. The problem here was that the entire part weighed less than half a gram. A good melt viscosity determination requires at least 20 times that amount. We were forced to turn to more sophisticated tests that could work with small quantities of material.

The first step was to compare the chemical state of the failed part with an as-molded part and the raw material. We did this by infrared spectroscopy. Figure 2 shows the results of these scans. The raw material and the molded part give an excellent match, but the failed part shows a significant problem.

At 1740 cm-1 there is a very strong absorption peak that is associated with groups that form when a polymer like nylon oxidizes. Some of these groups are present naturally in nylon materials, but clearly the incidence of these groups had increased dramatically. In addition, we saw changes in the spectrum that indicated that the original polymer chains were being broken and that there were now many short chains where once there had been a few long chains. Of course, all of this spells trouble for the polymer.

Solution Viscosity Test
While the infrared detects the problem, it does not quantify it. For this we still needed a viscosity test. When only small amounts of material are available, the test of choice is a solution viscosity test called an inherent or an intrinsic viscosity (IV) test. This measurement may be familiar to processors that work with bottle-grade PET polyesters. A full description of the principles and procedures behind the IV test are well beyond the scope of this article. However, in brief the technique involves comparing the viscosity of a known amount of the polymer dissolved in a solvent with the viscosity of the pure solvent. As with all viscosity tests, the higher the viscosity of the solution, the higher the molecular weight of the dissolved polymer. One of the big advantages of this test is that it can be run with a very small amount of material.

As with the melt viscosity tests that we have discussed in previous articles, there are rules governing acceptable and unacceptable changes in intrinsic viscosity. When comparing pellets to parts, the IV should not decline by more than 10 percent if the process is sound. Above 10 percent properties will begin to drop off, although good part design can add a safety factor and permit parts with 15 to 20 percent decreases to survive. In this case we compared pellets, the failed part, and as-molded parts. We wanted to be sure that the effects of processing were isolated from those of field exposure.

The as-molded parts showed a reduction of less than 7 percent from the IV of the pellets. This established the process as capable in terms of producing parts with good property retention. The failed part, on the other hand, showed a shift of more than 13 percent. This put an exact number on the qualitative data provided by the infrared. We now knew that field exposure was chemically changing the polymer and was reducing its average molecular weight in the process.

Signs of Crosslinking
But there was more. During the process of dissolving the various samples, the technician working on the IV tests noted that a substantial portion of the failed sample did not dissolve readily. Given the strength of the solvent we were using, this was remarkable. Typically, failure of a polymer to dissolve is a sign that crosslinking has occurred. We normally associate crosslinking with thermoset materials. In these systems a network solid forms that is very resistant to heat and solvents. This network also prevents the system from remelting once it has formed.

Crosslinked materials are very rigid, but they also possess limited elongation and are therefore extremely brittle for the most part. Certain thermoplastic polymers are also capable of crosslinking in a limited manner, and often this occurs in the advanced stages of molecular weight reduction. Reactive sites on the chains of reduced size begin to link together. In fact, if you look carefully at Figure 1, you will notice that the material aged at 120C had actually started to come back down in melt-flow rate. This can be an early sign of crosslinking.

If carried to extremes, it is possible that much of the initial change in melt-flow rate can be erased. Why is this important? Because as these crosslinks begin to form, the viscosity test reads this as an increase in molecular weight without giving any consideration to the fact that it is the wrong type of molecular weight increase. If this had happened to our part, then the reduction of 13 percent could represent a more drastic reduction in chain length followed by a competing buildup in molecular weight through crosslinks. If this were occurring, it would shed new light on the severity of the test bed conditions.

To test for this possibility, it was necessary to repeat the IV test and subject the solution to a more sophisticated test known as gel permeation chromatography (GPC). This is an expensive and time-consuming test, and for this reason it is seldom used in standard problem solving. But a lot was at stake in this case, and it was the only viable method for examining the anatomy of the IV results.

Again, the principles of GPC are complicated, but if you have worked with SPC, you can understand the relationship between IV and GPC. IV is a reflection of the average molecular weight of the polymer sample, while GPC provides a histogram of all of the individual chain sizes that make up that average. Obtaining an IV number is like getting a mean dimension on a sample population, but the GPC is analogous to seeing the entire distribution of measurements that make up that average.

Figure 3 shows the results of the GPC for the as-molded part and the failed field material. This confirmed our suspicions. The field failure had a wider distribution, proving that while some chains had been reduced in size during initial degradation, some of these chains were now joining up to produce a higher molecular weight fraction as well. The average effect on the IV tended to cancel out to a point, but the real problem was much worse than the 13 percent test value suggested. Here was a case of chain scission followed by oxidative crosslinking, a process that required considerable heat to initiate. This told us that the temperature of the test bed was well beyond the stated 110C.

Cons of Accelerated Testing
A review of property retention data as a function of temperature revealed the problem, and it is a common problem with a lot of accelerated testing. The material that had been selected was designed for long-term exposure at temperatures up to 120C. Up to that point, the polymer and the stabilizers in the compound were designed to protect the material from the harmful effects of oxidation. Test data from the material supplier showed that after 5000 hours at 120C property retention was essentially 100 percent. But above this point deterioration begins to set in quickly.

Degradation processes such as oxidation tend to proceed exponentially according to a rule of thumb that says that for every increase of 10 deg C a particular level of degradation will be achieved in half of the time. This means that an increase of 20 deg C shortens the life of the product by a factor of four, a 30 deg C rise cuts it by a factor of eight, and so on. But below the threshold temperature, the degradation will not occur to any appreciable degree. This is the fallacy of some accelerated tests. They venture into territory where they provoke changes in a material that could not occur in the actual application. They are predictors of catastrophic failure, not true accelerated testing.

Armed with this collection of test data and property information from the material supplier, our client was able to go back to its very imposing customer and explain that the material was properly molded and was designed to handle the application conditions as they were originally specified. In this case, the material was doing exactly what it was supposed to do. Exposed to temperature extremes beyond its limits, it was undergoing physical and chemical changes that rapidly caused the product to deteriorate. This had led to the field embrittlement. A little time spent evaluating and understanding the fundamental behavior of the material saved the molder, and ultimately the end user, many thousands of hours of continued testing and failure analysis. It also placed the responsibility firmly with the end user to define better the conditions under which it expected the material to operate.

Sign up for the PlasticsToday NewsFeed newsletter.

You May Also Like