The Materials Analyst, Part 52: Where does the moisture go? (Part 2)

Editor's note: This month's Materials Analyst is the second part in a series on measuring moisture content in resin. In the June issue of IMM we will publish a rebuttal to some of the arguments stipulated by Michael Sepe regarding the accuracy and effectiveness of loss-in-weight moisture analyzers.

In our last article, we began a discussion of the effects of moisture on polymers during melt processing and outlined the importance of being able to accurately measure the moisture content of a material. We also began to deal with the different approaches to moisture measurement on the market and the importance of method development in arriving at the correct value when a moisture test is performed. In this segment, we will discuss in greater detail the different methods of measurement that are available to molders and point out the advantages and pitfalls that come with each of them. Once we have established a scientific method of measuring moisture content, we can then start the discussion of where the moisture goes as the pellets travel through the molding process.

Moisture-specific Analysis
As we mentioned in our last article, the methods for measuring moisture that are recognized by ASTM all involve a chemical reaction that ensures that the only thing being measured is water. Most of these methods utilize a method known as Karl Fischer titration. This sounds intimidating, but an adaptation of the instrument that allows moisture to be extracted from the plastic in a remote location, combined with the wonders of the microprocessor, has turned this into a very manageable technique.

The method that is most commonly used conveys volatiles from a heated sample of polymer into a chemical solution. The water produces a chemical reaction that changes the electrical conductivity of the solution and an electrode records this change and converts it into a calculation of collected moisture.

The benefit of chemical techniques such as Karl Fischer is that they are moisture specific. A number of chemicals may be swept out of the heated pellets, but ideally the only substance that registers in the test is water. As we mentioned last time, method development is important.

First, it is essential that the polymer sample not degrade during the test. Some degradation byproducts can react with the chemicals in the titration vessel to produce water or create chemicals that interfere with accurate moisture measurement. PET polyester is one that was mentioned in the last article; acetals can pose the same problem. PPO materials have been known to give impossibly high moisture content values when run at elevated temperatures in air. Materials that contain halogenated flame-retardant additives will also pose problems. The best way to extend the acceptable range of test temperatures is to run the tests in a relatively inert atmosphere such as nitrogen or argon. Any instrument that does not permit the use of an oxygen-free gas stream is suspect.

A second problem, and one that the use of an inert atmosphere will not prevent, is a phenomenon known as solid-state polymerization. This problem is particularly common in nylon materials. When a material like nylon 6/6 is synthesized, the reaction is known as a condensation reaction and water is produced as a byproduct. This water is removed from the system during synthesis and finishing.

However, if the polymer is heated to very high temperatures,

a small amount of additional polymerization may occur, particularly in materials that are very dry. This reaction can occur at temperatures where the material has not melted, hence the name solid-state polymerization, or solid phasing. If this occurs, new water will be created and it will register in the moisture content test as though it were part of the original sample.

ASTM D 789, which deals primarily with nylon materials, shows a method development curve where the measured moisture content rises with increasing temperature, levels off into a plateau, and then rises again. This second rise may be due to either degradation or solid-state polymerization. In nylons, it is usually the latter, and it is therefore important to keep the measurement temperature in the plateau region.

Shop-floor Ingenuity
While methods like Karl Fischer are recognized techniques, moisture measurement takes many forms on the shop floor. The most rudimentary method involves either examining parts for cosmetic signs of moisture like splay and silver streaking or purging a shot into the air and looking for bubbles and froth in the air shot. Many molders who actually invest in a moisture measuring instrument are surprised to find that only about half of the cosmetic problems that appear to be caused by excess moisture actually involve wet materials. Worse yet, some materials like PET polyester give no cosmetic clues that they are being processed with excess moisture.

A small step up from the purge-and-look technique is the glass slide and hot plate method. This involves heating a glass slide on a hot plate on which several pellets of material have been placed. Once the pellets have softened, a second slide is placed on top of the pellets and they are pressed into a thin film. The cooled press-out is then examined for bubbles. This method gained considerable credibility in the 1960s when it was dubbed by one material supplier as the Tomasetti Volatile Indicator (TVI) test. Who is going to argue with that? This test actually worked fairly well for many unfilled materials, particularly PPO-based alloys. It was far less successful for filled systems in which air can become trapped between the polymer and the filler and produce bubbles. The test did what it was advertised to do—it detected volatiles, including but not limited to water. And it was most certainly not quantitative.

Measuring More Than Water
Given this level of sophistication, it is understandable that most molders who take the time to monitor moisture use a loss-in-weight system and consider it a huge leap forward. But these systems have two problems. One is that most of them run the test in ambient air. The second problem is that they assume that all the mass lost during the test is due to water. Unfortunately, a lot of other substances such as lubricants, other additives, and residual monomers can also be driven off, and there is no way to distinguish between the various substances in the weight loss fraction. This is particularly true if the test is run at a temperature high enough to remove all of the water. If loss-in-weight moisture tests are run at an appropriate temperature, they will almost always produce a result that is higher than the actual moisture content.

This leaves builders and users of loss-in-weight systems with a difficult situation. Even dry material tests as though it were wet. You may have noticed that as the loss-in-weight system manufacturers have become more sophisticated, they have begun to check their instruments against—you guessed it—Karl Fischer. Manufacturers of loss-in-weight systems even publish lists of moisture measurements made using their system and the Karl Fischer system in an effort to show how well their instruments work.

But there is a problem. At a given test temperature, a Karl Fischer test and a loss-in-weight system will produce similar reductions in sample mass. But assuming that no polymer degradation has occurred, the Karl Fischer system will only count that part of the weight loss that is water, while the loss-in-weight system counts everything. To get the loss-in-weight system to provide the correct values, the test temperature is reduced until the answer agrees with the Karl Fischer determination. It is important to understand that by reducing the test temperature we have not ensured a moisture-specific measurement; we have simply reduced the sum total of all the collected gases so that it agrees with the actual moisture content measured by the chemical technique. This can be made to work, but it is an exercise that must be conducted one grade of material at a time and a Karl Fischer technique or an equivalent must be available to provide the final word on measurement accuracy.

Litmus Tests for Analyzers
There are some litmus tests that should be run on an instrument to check its method of operation. First, as we noted, the temperature region where the moisture measurements for a given material are not influenced by temperature is typically 20 to 30C. A measurement system that truly removes all of the moisture in the material will produce a sample at the end of the test that has been completely dried, as long as an actual collection rate endpoint is used to stop the test. If you want to check a moisture measurement system's principle of operation, run a moisture test at the desired temperature and make sure that the endpoint of the test is triggered by a collection rate of zero rather than an extrapolated endpoint based on the first few minutes of the test. Allow the sample to cool and then run the test again using a test temperature 10 to 15 deg C higher than that used in the first test. If the first test has really removed all of the water, then the result of the second test should be zero. Loss-in-weight systems that run at an inappropriately low temperature in order to suppress excess weight loss will not meet this criterion, particularly if the tests are run in air.

Second, most moisture measurement devices, regardless of their fundamentals, can use the microprocessor to predict the outcome of the test based on the first few minutes of the actual test. This is designed to save time, but it is possible to select conditions for the extrapolation that can introduce error. If you plan to use the "predict" function instead of the "actual" measurement, be sure that the two methods give the same answer.

Third, allowable moisture content parameters provided by material suppliers are usually given in terms of percent moisture. Many of the more sensitive materials have upper moisture content limits in the hundredths of a percent. If the system you are evaluating measures in percent, be sure it has a working third decimal place. Instruments that go directly from .02 percent to .03 percent give a result of .02 percent for materials that contain up to .025 percent moisture. For some materials like PET polyester and polyurethane, this is already high enough to cause problems. If your system measures in parts per million, make sure that it provides a significant figure in at least the "tens" column so that it can distinguish between 210 and 220 ppm (.021 percent and .022 percent).

Don't Get Too Simple
Finally, the results coming from the instruments should be reasonably repeatable, and they must agree with the reality of the process. The repeatability aspect need not be at the analytical laboratory level. For processors, the moisture measurement is designed to distinguish between wet and dry material. While .012 percent to .016 percent may represent a 33 percent variation, the conclusion is the same. But an instrument that reads .02 percent one time and then .06 percent the next is of

no use in plastics processing. If a material checks in the moisture analyzer as dry, and repeated attempts to process encounter obvious signs of wet material, there may be a problem with the instrument. Many times processors are baffled by obvious problems with processing because they have a moisture monitor that tells them the material is dry, and they believe it unwaveringly.

Einstein said that things should be made as simple as possible and no simpler. He could have been talking about moisture analysis in the plastics industry. The reality is that the loss-in-weight techniques that many processors have adopted were developed for industries in which the accuracy of the measurement is acceptable in terms of tenths of a percent such as the food or textile industry. For materials where the difference between 180 ppm and 240 ppm is the difference between good parts and bad parts, and where a host of other substances can be driven off during the test, they may be simply inadequate.

Accurate Solutions
There are two alternatives. One is the chemical approach. A lot of processors are hesitant to work with an instrument that looks like a piece of laboratory apparatus. There is glassware, tubing, and chemicals. But once these instruments are set up and the methods are developed, they are surprisingly user friendly and the annual operating costs can be less than $1000 even for an instrument that sees frequent use.

If these types of instruments still look too scary, the other option is a system with a moisture-specific sensor. These instruments drive off volatiles just like any other system, but they monitor the gas stream and recognize only the water. These systems are more hardened for the production floor, but in order to be accurate, they must still be kept in a controlled atmosphere. They can also be influenced by changes in humidity and temperature, just as a Karl Fischer system will. The moisture-specific systems eliminate the chemicals, but they may have other disposables that actually cost more. And in the final analysis, these instruments still must give results that match those of the time-honored chemical methods.

In the end, an inaccurate instrument is almost worse than no instrument. Processors who don't check the material for moisture at least know that they don't know. But those with substandard instruments believe that they know, write procedures around the use of the instrument, and might still make bad parts. So if it's worth the time and money to do these measurements—and it certainly is—then it pays to use an instrument based on sound scientific principles. These are the principles we will use next time to track our moisture as it travels with our pellets through the molding process.