The Materials Analyst, Part 119: The science and mythology of 
measuring moisture content in plastic 
materials—Part 3

Will the day come when an accurate plug-and-play moisture analyzer is available to the masses? Not without an acknowledgement of the problems with our current systems.

Let us assume for a moment that you have conducted the experiments on your loss-on-drying (LOD) system that were outlined in Part 2 of this series, and the instrument has failed to provide any indication that it is capable of extracting and measuring a fixed mass of volatile material. This alone would appear to be a major concern regarding the accuracy of the instrument.

But countering this concern is the claim made by the manufacturers of LOD systems that the instruments are calibrated. Calibration presumes that there is a standard that is used to perform this exercise. And this is where it gets interesting and, for many, confusing.

There are materials sold that contain a particular amount of moisture and are used for the purpose of calibrating moisture analyzers. Typically these are hydrated minerals, inorganic compounds that have a known amount of water bound into the crystal structure. At an appropriate temperature the bonds holding the water to the mineral-based crystal structure will be broken and the water can be removed.

These materials can be combined with inert, hydrophobic ingredients in order to produce mixtures that are designed to target particular moisture content ranges such as 0.01%-0.10% or 0.10%-1.00%. These compounds, when tested properly in an LOD system, will give good agreement to the standard values. This should be expected because these materials fit the description provided in Part 1 of this series for materials that work well in an LOD instrument. They contain a relatively large amount of water and they contain no other substances that are volatile enough to compete with the evolved moisture.

Unfortunately, these calibration compounds are not at all like the plastic materials that processors wish to evaluate. As we have already discussed, polymers contain a host of other components. Many of these evolve in the same temperature range as the moisture that comes out of the polymer. To answer these concerns, the manufacturers of LOD systems have come up with another type of exercise that they also refer to as calibration.

Let’s just drop that temp down a little . . .
Here is how it is done. A sample of resin is first tested using a Karl Fischer system. It may come as a surprise that all of the makers of LOD systems have Karl Fischer devices in their facilities. The reason is simple. Karl Fischer tests are needed to provide the correct answer regarding the amount of moisture in the sample.

Karl Fischer reagents are designed to detect the water that is extracted from a sample using a chemical reaction with iodine, one of the ingredients in the Karl Fischer reagents. The consumption of iodine creates an imbalance in the equilibrium between iodine and iodide ion and initiates the generation of iodine. Since the iodine and the water react in a 1:1 ratio, and the iodine consumed is proportional to the electrical current generated, the amount of water can be calculated using Faraday’s Law.

An electrode provides the measurement of the conductivity of the solution. This technique is known as coulometric titration. Today’s instruments turn the calculation over to a microprocessor so that the operator simply sees a display tabulating the collected moisture in real time. Plugging in a sample mass produces the moisture content in percent or parts per million.
Because some polymers contain a variety of additives and may begin to degrade at the temperatures required to efficiently extract moisture, there are precautions that must be taken, even with Karl Fischer apparatus, in order to obtain an accurate result. We will come back to some of those a little later in this article. But for now, let us assume that the Karl Fischer test has been performed properly and the moisture content of the sample has been accurately measured.

The next step in the so-called calibration involves taking a sample of the same material and placing it into the LOD apparatus. Temperature and time are critical to ensuring a complete extraction of all the water in the sample, and the method for determining the correct temperature was reviewed in Part 2 of this series. We will come back to the methods for determining the correct time.

Let us assume that the material being tested is polycarbonate (PC). In a Karl Fischer apparatus a test temperature of 190°C-220°C will provide for rapid and complete extraction of all the moisture in the sample and 200°C is a commonly used temperature that falls in the middle of this range.

If the PC is tested in the LOD system at this temperature, the mass of material evolved will include moisture, additives, and residuals that will weigh far more than just the water that was measured in the Karl Fischer test. Of course, the Karl Fischer test also drove off these additional constituents, but the instrument counted only the water and ignored the other ingredients.

The LOD system is not so discriminating. Since it operates based on weight loss, it will count the entire change in the weight of the sample as moisture. So if the PC sample is tested at 200°C, the LOD system will report a number much higher than that obtained from the Karl Fischer titrator. No one would ever start up their molding machines, because the samples would continue to produce values well above the maximum value of 0.020%, regardless of how long the material was dried.

The manufacturers of the LOD systems get around this problem simply by reducing the test temperature of their apparatus until the number obtained by the LOD instrument agrees with the value obtained by the Karl Fischer system. Incredibly, this is what is referred to as calibration.

Avoiding the quacks
Typically, the test temperature used by the LOD system will be 50-75 deg C lower than the temperature used in the Karl Fischer titrator. This reduction in temperature ensures that some of the water that was in the pellets is left behind, but the additional components evolved make up the difference. It is tempting to allow for this “correction” factor until we consider that this equivalence will only work as long as the concentration of all the other components remains exactly the same. Since this will never be the case in the real world, the equivalence is not a useful guideline, much less something that rises to the level of a calibration.

This would be like going to the doctor for a cholesterol test and finding out that the instrument being used to measure your cholesterol cannot distinguish between cholesterol, triglycerides, blood glucose, and so forth. It will count everything it sees as cholesterol. So the technician running the instrument, who happens to have your results from the last time you were in, scales down the sensitivity of the instrument so that it produces a number similar to the one that was obtained previously.

It is not really a measure of cholesterol. It is an assay of a large variety of substances measured in such a way as to agree with a previous measurement. Once we understand this sleight of hand it is a little easier to appreciate why Dr. Ezrin was so incredulous when I informed him of how moisture measurements were being done in the plastics industry.

The reality is that in order to use any test technique effectively, it is important to understand the method of measurement and the various influences that can interfere with the accuracy of the technique. When we first adapted Karl Fischer titration for plastic materials, we found that the additives and degradation byproducts evolved by heating could interfere with the chemistry of the Karl Fischer system in a unique way.

It is important to remember that accurate moisture measurement involves complete removal of the water in the sample without allowing the measurement system to detect any other ingredients. But initially all of our tests were performed using dry air as the medium for removing the moisture from the samples. We found that by the time we reached a temperature high enough to remove all the water, we had also started to evolve other ingredients that gave us nonsensical results. PPO materials, for example, which cannot hold more than 0.07% water, were producing results almost 10 times higher.

Particularly puzzling was the test result obtained for PET polyester. PET samples would run indefinitely without the instrument achieving an end point. We discovered that when PET begins to degrade, it produces a chemical known as acetaldehyde or AA. Bottle manufacturers are very familiar with this phenomenon because AA has a slightly sweet taste than can influence the flavor of the beverage. AA levels are tightly controlled by limiting the upper end of the melt temperature profile for molding the preforms and subsequently blowing the bottles.

One of the chemicals in the Karl Fischer reagents is methanol. As it turns out, aldehydes react with alcohols to form—you guessed it—water. We were manufacturing water during the test! However, we found that by substituting dry nitrogen for dry air, we could open up the temperature window between the point needed to remove all the water and the point where polymer degradation began. This brings us to another flaw in the LOD systems. They do not have the ability to control the test atmosphere; they run on ambient air.

Keep it simple, 
but not too simple
Yet another instrument parameter that can be manipulated is the endpoint determination. This can be done with almost any instrument, and the objective is to shorten the test time by using an algorithm to predict the remaining amount of moisture that will come out of the sample based on the initial portion of the test result.

This is another method that can be used to adjust the values obtained by an instrument so that they “agree” with more rigorous methods. The best techniques employ an endpoint that represents the actual end of the moisture evolution process. However, this is very difficult to do with an LOD system because of the large amounts of other ingredients that are being released along with the moisture.

Einstein once commented, “Everything should be made as simple as possible, but not simpler.” LOD moisture analysis for polymeric materials is an example of something that has been made too simple. The methods are easy, but they do not provide reliable accuracy. When they do provide good agreement with accepted methods, it is literally luck. Yet there is a need for an instrument that is moisture specific and employs the ease of use demonstrated by an LOD system.

Work on such a sensor-based device has been going on for some time now, shows great promise, and is actually commercial. But it is more expensive than either an LOD or a Karl Fischer and method development by the user will still be needed. We have not yet arrived at the plug-and-play stage when it comes to moisture analysis, although we could get there someday if we are willing to face the limitations of our current methods and work hard.

But continuing to ignore the fundamentals and pretend that, because an instrument can generate a number, it represents an accurate value when the method of measurement is fundamentally flawed serves neither the processor nor its customers. The fact that representatives of some resin suppliers have actually started to endorse LOD instruments simply furthers the departure from good science and encourages hazy thinking and sloppy procedures. We can do better, and given the fact that polymer degradation remains a major problem in plastic part failure, the incentives to be better are high.

You can read Part 2 of this article here, and Part 1 here.