3.13 Ageing tests
Main page |
Ageing tests are one of those things in product development where things can go really wrong. You want to make sure that the product stays within specification at least in the specified product life time/shelf life.
The one test you always have to do is the real time test, and in most cases it's plural, tests. It is the reference point for all other tests, and it is one of mandatory tests in product development. What you are looking for is ageing phenomena.
Real time means having the product in an environment like the one where it is used for as long as the product is supposed to last. If it's a product for use in hospitals, you use room temperature in an indoor environment. If it's paint for outdoor use, you paint a piece of wood, metal, concrete or whatever the paint is for and place the sample outside in the environment where it is supposed to be used.
Simple yet complicated because you have to do a number of tests. We'll get back to those.
But you can't wait 3, 5, or 10 years for a real time test before product release, can you? That would make no sense. For this you make an accelerated ageing test, and this is where things can get a little tricky.
Accelerated ageing is done by exposing the material to the same conditions as the real-time ageing, only at elevated temperatures. According to Arrhenius' Rule, reaction speed is doubled for every 10 °C the temperature is increased, so by increasing the temperature 10 °C, you can double the ageing speed. This part most people can understand, but this is also where people make the worst blunders.
For some reason, the understanding of the phenomenon stops at "doubling the speed when increasing the temperature 10 °C". This is only a model that works for simple systems, and it only works within a narrow temperature range.
Consider an ordinary chicken egg. If it is placed at room temperature for 80 days, it'll rot. 21 days under a hen (37 °C) will produce a chicken. 8 hours in boiling water gives you a hard-boiled egg. According to Arrhenius you should get the same result.
So what goes wrong?
The doubling per 10 °C is a piece of math, and the model only works for very simple systems or very narrow temperature intervals. The equation looks like this:
R is a natural constant, and k0 and Ea are reaction specific constants. The doubling is the math from the exponential function where the temperature T becoming (T + 10 °C).
So what does that mean? R can't be changed, no matter what. That's the nature of natural constants. k0 and Ea are reaction specific constants. That means that for this specific reaction they can't be changed either. If they change, then your reaction has changed, which is exactly what happens with the egg. You are no longer measuring the same reaction/ageing as your real time test and your accelerated ageing test is a lie.
The math in the equation is also the reason why you can't have something like "an almost fit" or apply some other correlation like double reaction time per 9 °C. Another variation of this is that "we know that something happens to the product while ageing that we don't see at real time, but if it can stand the accelerated ageing, it's safe to assume the real time ageing will also be okay". Two things: The chemistry involved in ageing doesn't work that way, the higher temperature may have stabilized the product instead, and you have just made yourself guilty of forgery.
Don't EVER do that!
Here is how it's done:
For complex systems there are limitations to how much the temperature can be increased and still get useful results. The temperature limit is material specific. If the limit is not known, a series of parallel experiments at various temperatures should be set up. The results of the accelerated tests are correlated to the real-time test, so the result for 1 month at 40 °C should be the same as 2 month at 30 °C and 4 month at 20 °C.
The validity of the results of the accelerated test is really not known until the real-time test reaches the same point in time. Until then it is only a guess. The longer the real-time and accelerated test correlates, the more qualified the guess, but until real-time data are available, you really don't know how good the guess is.
When reaching the point where the accelerated test deviates from the real-time experiment, you stop the experiment. If you continued the experiment you risk abuse by someone who "just want a number" and don't care if it's right or wrong. Also what would be the point of continuing? The experiment was set up to predict ageing phenomena and you know by now that what you get from the measurements is not a prediction. Do you need the extra work load? Focus on the experiments that give you the info you need, and stop wasting time.
A variation of this is ageing composite products like adhesive and backing or adhesive and release liner. The highest temperature at which all parts shows accelerated ageing according to Arrhenius equation is the maximum temperature for ageing the composite. If you are only interested in, say the stability of the release liner on an adhesive (e.g. a sticker or a band-aid), it is compelling to do the ageing test at the highest temperature this component can take, even if the adhesive will be ruined. This is a really bad idea. When a product starts to degrade, a chemical reaction takes place, small mobile molecules are formed, and in the contact area this is bound to affect the material in contact with the degradation products by introducing unknown side reactions. Also, some of the degradation products are bound to be small mobile molecules, migrate from the adhesive to the other components will occur, changing their properties. It may be nothing, but it may as well be everything, you have changed the chemistry, so you are doing the measurements on something other than the material you want to test.
Many changes in material properties as a function of ageing are fastest in the beginning. It is therefore not common to make the measurements at regular intervals throughout the study. Instead a lot of measurements are made in the beginning, and then the intervals are gradually increased to a suitable size. A typical ageing test over 3 years would be start, 1 week, 2 weeks, 1 month, 3 months, 6 months and after that, every 6 months. Just like the frequency of the measurements decrease, the number of measurements will often decrease too. Especially in screening experiments where some of the tested materials can be expected to fail.
As experience grows with a material, the ability to make a qualified prediction on ageing conditions and effects will grow too. Therefore one should expect the first ageing tests of a new material to require quite a lot more work than the following e.g. the first adhesive of a given type is tested at two elevated temperatures and as the batch deviation is unknown five products are tested every time as recommended by FDA. Later as adjustments are made on the adhesive formulation, only one elevated temperature is needed, and if the batch variations are sufficiently low, only three products are needed for each test.
The one test you always have to do is the real time test, and in most cases it's plural, tests. It is the reference point for all other tests, and it is one of mandatory tests in product development. What you are looking for is ageing phenomena.
Real time means having the product in an environment like the one where it is used for as long as the product is supposed to last. If it's a product for use in hospitals, you use room temperature in an indoor environment. If it's paint for outdoor use, you paint a piece of wood, metal, concrete or whatever the paint is for and place the sample outside in the environment where it is supposed to be used.
Simple yet complicated because you have to do a number of tests. We'll get back to those.
But you can't wait 3, 5, or 10 years for a real time test before product release, can you? That would make no sense. For this you make an accelerated ageing test, and this is where things can get a little tricky.
Accelerated ageing is done by exposing the material to the same conditions as the real-time ageing, only at elevated temperatures. According to Arrhenius' Rule, reaction speed is doubled for every 10 °C the temperature is increased, so by increasing the temperature 10 °C, you can double the ageing speed. This part most people can understand, but this is also where people make the worst blunders.
For some reason, the understanding of the phenomenon stops at "doubling the speed when increasing the temperature 10 °C". This is only a model that works for simple systems, and it only works within a narrow temperature range.
Consider an ordinary chicken egg. If it is placed at room temperature for 80 days, it'll rot. 21 days under a hen (37 °C) will produce a chicken. 8 hours in boiling water gives you a hard-boiled egg. According to Arrhenius you should get the same result.
So what goes wrong?
The doubling per 10 °C is a piece of math, and the model only works for very simple systems or very narrow temperature intervals. The equation looks like this:
| -Ea | |
| k = k0 · 10 | R·T |
R is a natural constant, and k0 and Ea are reaction specific constants. The doubling is the math from the exponential function where the temperature T becoming (T + 10 °C).
So what does that mean? R can't be changed, no matter what. That's the nature of natural constants. k0 and Ea are reaction specific constants. That means that for this specific reaction they can't be changed either. If they change, then your reaction has changed, which is exactly what happens with the egg. You are no longer measuring the same reaction/ageing as your real time test and your accelerated ageing test is a lie.
The math in the equation is also the reason why you can't have something like "an almost fit" or apply some other correlation like double reaction time per 9 °C. Another variation of this is that "we know that something happens to the product while ageing that we don't see at real time, but if it can stand the accelerated ageing, it's safe to assume the real time ageing will also be okay". Two things: The chemistry involved in ageing doesn't work that way, the higher temperature may have stabilized the product instead, and you have just made yourself guilty of forgery.
Don't EVER do that!
Here is how it's done:
For complex systems there are limitations to how much the temperature can be increased and still get useful results. The temperature limit is material specific. If the limit is not known, a series of parallel experiments at various temperatures should be set up. The results of the accelerated tests are correlated to the real-time test, so the result for 1 month at 40 °C should be the same as 2 month at 30 °C and 4 month at 20 °C.
The validity of the results of the accelerated test is really not known until the real-time test reaches the same point in time. Until then it is only a guess. The longer the real-time and accelerated test correlates, the more qualified the guess, but until real-time data are available, you really don't know how good the guess is.
When reaching the point where the accelerated test deviates from the real-time experiment, you stop the experiment. If you continued the experiment you risk abuse by someone who "just want a number" and don't care if it's right or wrong. Also what would be the point of continuing? The experiment was set up to predict ageing phenomena and you know by now that what you get from the measurements is not a prediction. Do you need the extra work load? Focus on the experiments that give you the info you need, and stop wasting time.
A variation of this is ageing composite products like adhesive and backing or adhesive and release liner. The highest temperature at which all parts shows accelerated ageing according to Arrhenius equation is the maximum temperature for ageing the composite. If you are only interested in, say the stability of the release liner on an adhesive (e.g. a sticker or a band-aid), it is compelling to do the ageing test at the highest temperature this component can take, even if the adhesive will be ruined. This is a really bad idea. When a product starts to degrade, a chemical reaction takes place, small mobile molecules are formed, and in the contact area this is bound to affect the material in contact with the degradation products by introducing unknown side reactions. Also, some of the degradation products are bound to be small mobile molecules, migrate from the adhesive to the other components will occur, changing their properties. It may be nothing, but it may as well be everything, you have changed the chemistry, so you are doing the measurements on something other than the material you want to test.
Many changes in material properties as a function of ageing are fastest in the beginning. It is therefore not common to make the measurements at regular intervals throughout the study. Instead a lot of measurements are made in the beginning, and then the intervals are gradually increased to a suitable size. A typical ageing test over 3 years would be start, 1 week, 2 weeks, 1 month, 3 months, 6 months and after that, every 6 months. Just like the frequency of the measurements decrease, the number of measurements will often decrease too. Especially in screening experiments where some of the tested materials can be expected to fail.
As experience grows with a material, the ability to make a qualified prediction on ageing conditions and effects will grow too. Therefore one should expect the first ageing tests of a new material to require quite a lot more work than the following e.g. the first adhesive of a given type is tested at two elevated temperatures and as the batch deviation is unknown five products are tested every time as recommended by FDA. Later as adjustments are made on the adhesive formulation, only one elevated temperature is needed, and if the batch variations are sufficiently low, only three products are needed for each test.