Once again, a reduction in MFR is not always good for your material. Pay attention to the polymer chains and beware too much of a good thing.
How will the automotive underhood market fare this year? Hold on to your engines, because it’s going to be a bumpy ride. Last month we discussed some instances in which the melt-flow rate (MFR) for molded parts can be lower than that of the base resin from which the parts are produced. We showed that large downward changes in MFR can be just as damaging to part performance as large upward changes, but the causes for these downward trends are more complicated and often involve competing processes.
This month we will look at other materials that can react in unexpected ways when particular ingredients are present. These materials can also produce MFR results that may appear on the surface to be acceptable, but in reality spell trouble and cause parts to fail.
Chain extenders
The actual product failure that first revealed this problem in our lab involved large parts manufactured from a 30% glass-fiber-reinforced PET polyester. The polymer being used in the filled compound was a recycled material. To prevent degradation during compounding and the subsequent molding process, the manufacturer of the material incorporated a chain extender into the formulation.
A chain extender is designed to counter the effects of chain scission by allowing small chain segments to link up to form larger chains.
Ideally, the chain extender is incorporated at a level that will just balance the tendency for the chain scission that may occur due to thermal and hydrolytic degradation during melt processing. However, variations in drying and processing conditions make this balance point very difficult to gauge, so the tendency is to provide a safety factor by adding an excess of the chain extender.
In theory, this sounds like a good idea. However, excess chain extender will continue to work on the polymer during the molding process. One of the practical consequences of this was that processors noted a tendency for the material to increase in melt viscosity, particularly at times when the process was interrupted. On restarting the machine, the material would exhibit a higher viscosity until the material in the barrel was cleared and new resin from the hopper introduced. Molders know that, generally, the material that exits the nozzle immediately after an extended process interruption flows more easily, so this behavior was immediately notable.
When parts began to come back from the field, they all showed signs of fatigue failure, a mechanism that was consistent with the manner in which the product was used. Usually, when PET polyester parts fail in the field, the immediate assumption is that the raw material was molded with excess moisture and the polymer degraded rapidly. Melt-flow tests often show catastrophic increases from pellets to parts that confirm this.
However, when these failed parts were tested, they showed almost no change in MFR compared to the raw material. In some cases the MFR of the molded part was actually lower than that of the pellets. This is remarkable because in a glass-fiber-reinforced material, even if the polymer comes through the process unchanged, there will almost always be an increase in melt flow simply due to the reduction in glass-fiber length.
Unexpected decrease in MFR
In a 30% glass-fiber-reinforced PET, the nominal melt flow for most commercial materials hovers around 10 g/10 min when tested at the prescribed conditions. A well-molded part will exhibit an MFR of 20-30 g/10 min, with this change representing the compound effect of a small decline in the average molecular weight of the polymer and a reduction in the length of the glass fibers.
A resin molded with elevated moisture content frequently shows increases into the range of 125-200 g/10 min. This indicates catastrophic levels of polymer degradation and the parts will be correspondingly brittle. However, in the case of this product, manifestly brittle parts were coming in with MFRs of 5-6 g/10 min. As with our examples from last month, this would appear to be a positive change and not one related to the brittle behavior of the molded product.
When MFR tests are conducted, the material placed in the heating cylinder of the instrument is allowed to come to equilibrium with the temperature of the chamber for approximately 6-7 minutes. A lot can be learned by comparing results obtained with this standard preheat time to those derived from purposely extending the preheat time. This allows for an evaluation of the material’s response to prolonged exposure to elevated temperatures.
Ideally, a material will display little or no change in MFR as a function of this longer dwell time. If a material exhibits poor thermal stability, the MFR will go up considerably. But in this case, the opposite trend occurred. As unusual as it was to observe the MFR declining during the molding process, extending the preheat time caused the values to drop even more.
The table captures the results of standard and extended preheat times for raw material and good and failed product. These results agreed with the observed increase in melt viscosity that the processor noted during process interruptions. The problem was how to reconcile what is typically considered to be a positive trend with the incidence of failed parts. To understand what is happening requires a more sophisticated level of testing.
A poor law of averages
When molecular weight is measured using a viscosity or MFR value, the resulting number represents an average. But in reality, commercial polymers are not made up of molecules that are all the same size. Instead, all polymers are a mixture of a wide range of molecular sizes. Some of the chains are relatively small, some very large, and most fall in a range close to the average value for the entire population.
The result is a distribution of molecular sizes or weights. When this distribution is plotted as a function of concentration, the result looks a lot like a histogram that might be obtained when a statistical analysis of a group of molded parts is conducted. Measurements of average molecular weight work well as long as we can assume that the molecular weight distribution is unchanged. But when the performance of the product contradicts the standard relationship with the molecular weight average, it is necessary to pull apart the material into its various chain sizes.
This is not that different from a situation in which a molder provides two shipments of parts and certifies that the average value for a critical dimension is the same in both batches of product. When the parts go to the assembly line, one lot of parts passes through without any problems but when working with the second lot, a number of parts fall out as being either too large or too small to fit properly. In a case like this, the quality engineer will begin to measure a larger sampling of parts in an effort to understand the distribution of sizes. Polymer chemists do the same thing, and the technique is known as gel permeation chromatography (GPC).
GPC involves dissolving a sample of the polymer in a solvent. In a filled material, this has the added benefit of removing the filler, which can interfere with an accurate interpretation of a value such as MFR. The solution is then injected into a column that contains a material known as packing. The material in this packing regulates the rate at which the polymer in solution flows through the column according to the size of the individual molecular chains. Large chains pass through most easily and therefore exit the column in the shortest time and are detected first. Smaller chains take longer. The resulting data is a sort of bell curve with the largest chains on the left hand side of the graph and the smallest ones on the right.
Figure 1 shows the result for a sample of our PET polyester raw material, a good molded part, and a failed molded part. The molecular weight distribution for the raw material and the good molded part reflect the typical curves obtained by GPC. There is very little change in the distribution between the raw material and the good molded part. This confirms that the effect of the molding process on the molecular weight of the polymer has been minimal.
However, the failed part shows a serious departure from the behavior of the raw material. Most of the polymer chains still reside in the middle of the distribution curve. Yet a significant population appears as a distinct peak on the left hand side of the plot. This represents a high-molecular-weight fraction. In addition, another small peak can be seen on the right-hand side, evidence that some smaller chains are also being created during the molding process. The net effect of this change is an increase in the average molecular weight, which corresponds to the observed decrease in MFR. But what is the nature of the newly created population?
Crosslinking corruption
To someone unaccustomed to looking at these types of curves, the occurrence of a peak just below 10 on the x-axis when the main peak appears near 13 may not seem like a problem. But the chains coming through the column in the left-most peak area are more than 10 times larger than those in the median of the population.
These chains do not represent simple chain lengthening; if they did, the product would be very impact resistant. Instead, they represent crosslinking, an unintended consequence of incorporating excess chain extender into the compound. Rather than consistently attaching to chain ends, the degraded fragments sometimes form branches off the side of the chains. When these link up to neighboring chains, the molecular weight increases rapidly and the crosslinked regions, while very stiff, are also not able to endure the deformation associated with the application and they break rather than bend.
The fact that this creation of crosslinked material is a dynamic process is shown in Figure 2. This plot compares the molecular weight distribution of two good molded parts to a sample of the material from the failed part that had been exposed to the extended preheat time in the MFR tester. Note that the entire molecular weight distribution is shifted to the left. However, the most important consequence of the prolonged heating period is a significant increase in the amount of crosslinked material. This shows that the ingredients placed into the compound are still active and are continuing to participate in a reaction that is detrimental to the function of the part. Imagine the impact of using the regrind from the process in making these parts.
True chain extension without crosslinking is possible, and it can also manifest as a decrease in MFR. However, when this occurs, the result is a true improvement in ductility. Figure 3 shows the effect that extended drying times have on the molecular weight distribution of a heat-stabilized nylon 6/6. In
this case, the material after drying had an MFR of 6 g/10 min while the material before drying had an MFR of 22 g/10 min.
The processor questioned whether this change was positive or detrimental. The answer is in the GPC result. In this case, the entire molecular weight distribution shifts in an approximately uniform manner to a higher average without the creation of any renegade species. And in fact, when the parts were molded from this lower-MFR material, they were extremely tough even before they were moisture conditioned.
Einstein once commented that things should be made as simple as possible, and no simpler. MFR tests are an appealing tool for assessing the average molecular weight of a polymer. But it must be remembered that we are measuring an average and that behind that average is a complex distribution of molecular sizes that can and do influence processing and performance. When the standard relationships between MFR and performance appear to break down, it is time to look deeper.
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.
The table captures the results of standard and extended 
Figure 1 shows the result for a sample of our PET polyester raw material, a good molded part, and a failed molded part. The molecular weight distribution for the raw material and the good molded part reflect the typical curves obtained by GPC. There is very little change in the distribution between the raw material and the good molded part. This confirms that the effect of the molding process on the molecular weight of the polymer has been minimal.
The fact that this creation of crosslinked material is a dynamic process is shown in Figure 2. This plot compares the molecular weight distribution of two good molded parts to a sample of the material from the failed part that had been exposed to the extended preheat time in the MFR tester. Note that the entire molecular weight distribution is shifted to the left. However, the most important consequence of the prolonged heating period is a significant increase in the amount of crosslinked material. This shows that the ingredients placed into the compound are still active and are continuing to participate in a reaction that is detrimental to the function of the part. Imagine the impact of using the regrind from the process in making these parts.
this case, the material after drying had an MFR of 6 g/10 min while the material before drying had an MFR of 22 g/10 min. 