Submitted by Randi Ersnt, FDR Design
We have been talking about using the MOCON OpTech to take oxygen
readings on units subjected to various hyper accelerated aging systems.
The thought being that due to the accuracy of the MOCON OpTech
instrument that we would be able to detect very small changes in the
argon fill of the test units.
About a year ago as part of a small group study seeking correlation of
gas fill readings a small task group worked on taking before and after
readings of test units undergoing the ASTM 2188/2189/2190 tests. First
phase was initial fill levels prior to any testing and we obtained good
correlation using three different MOCON OpTech readers measuring test
units that had the MOCON OpTech sensor placed in them prior to sealing.
When those units came out of ASTM testing we gathered again to repeat
the study. This time we found poor correlation between the SES and the
MOCON OpTech. A small number of units were taken to FDR and tested
using the FDR Gas Chromatograph (GC). We observed that the ratio of
Oxygen (O2) and Nitrogen (N2) in the "contaminate" within the sealed test
units was not the same ratio of O2 and N2 as found in ambient air.
Before drawing any conclusions we wanted to see a larger sampling of
units with diverse construction using a more precise GC testing method.
FDR recently obtained 18 units that were roughly one year old and had
undergone various stages of the ASTM tests.
Attached is a spreadsheet "FDR_GC_Aged IG Mocon" that shows the
results of those tests.
Again our purpose at FDR for taking the readings was to further explore if
the ratio of Oxygen and Nitrogen in aged sealed units was stable and
consistent with the ratio of O2 and N2 seen in air. We had tested a small
number of units as part of the task group study and had seen variability in
the ratio.
All of this pertains to the accuracy of test instruments in determining the
gas fill, specifically to Oxygen dependent instruments. The hope was with
the more accurate devices like the MOCON OpTech we would be able
to detect minute changes in the fill levels of units going through proposed
hyper accelerated tests.
What we saw confirmed what we saw the first time in that there is a large
variability in the ratio of O2/N2, we suspect the desiccant.
******
Spreadsheet notes:
In the spreadsheet the specific unit ID is an arbitrary random number 1-18.
Atmospheric air has 20.9% Oxygen and 78% Nitrogen (the missing % is
argon and other stuff) The ratio of O2/N2 is 3.73
the "FDR 90" is a certified manufactured bottle of gas we purchased that
FDR uses as a benchmark. "FDR 90" has 2.01% Oxygen, 7.66% Nitrogen,
balance is Argon Ratio is 3.81
The yellow tinted cells show what the average ratio was in the aged units
we tested (three samples) The ratio ranged from 1.90 to 5.12 (ratio in "air" is
3.73)
The red tinted cells show what the FDR GC says the argon content is (FDR
GC is set up with Argon carrier gas so it reads O2 and N2, we then assume
an Argon balance)
The blue tinted cells show what you would calculate the Argon content to
be based on measuring the oxygen and assuming a "normal" (3.73) ratio
of nitrogen and oxygen the same ratio as in ambient air.
Unit #18 for example is oxygen rich, it has a 86% argon fill but if you
calculated argon fill using oxygen it would be a 76% argon fill
Unit #10 goes the other way it has a 76% argon fill but using oxygen
calculates to be a 81% argon fill
Unit #5 is oxygen "rich" again so while it has a 91% argon fill using oxygen
calculates using oxygen to be a 85% fill
Three samples were taken from each sealed unit (photos attached). After
taking the 3 samples we went back to the known "FDR 90" bottle. So
sequence was three readings from cavity followed by one reading from
"FDR 90" bottle. Using a method of always returning to a known sample
from a bottle of certified gas.
******
Summary:
-The purpose of an inert gas fill is to improve the thermal performance of
the sealed unit.
-Oxygen and nitrogen the principle components of air have nearly
identical thermal conductivity so for thermal performance it does not
matter what the ratio of nitrogen to oxygen is in the remainder or
contaminant. Once a unit is filled and sealed the thermal performance of
the sealed unit is not significantly altered by a varying ratio of nitrogen
and oxygen.
-We assume that the ratio of nitrogen and oxygen is not a constant in an
aged sealed unit due to the dynamic nature of the desiccant and due to
the potential for gasses to move through the sealant at differing rates of
speed.
-Measuring a gas fill historically has been done assuming a ratio of
nitrogen and oxygen that is the same as that of ambient air.
-As oxygen is easier to measure than nitrogen or argon oxygen is often
measured then multiplied to determine the amount of nitrogen. Once the
assumed percentage of air is calculated it is subtracted from 100 to
determine the percentage of argon. While this method is attractive
because of the ease of measuring oxygen any variation in the oxygen to
nitrogen ratio creates inaccurate argon measurements. (attached
sketch)
-While researching the use of fluorescence oxygen sensing technology
(MOCON OpTech) we discovered that the ratio of oxygen to nitrogen in
aged units is not constant nor is it equivalent to ambient air. The
attraction of the oxygen sensing measurements was higher accuracy, this
accuracy could then be used to look for minor changes in the gas
fill. These minor changes could then be used to determine the integrity of
the seal.
-Oxygen sensing was proposed to be used in accelerated aging schemes
to measure minor seal leakage, FDR's Thumper - accelerated pressure
changes, NREL's Big Hammer - accelerated temperature swings, or ATI's
Oscar - accelerated P1 chamber. If the test units contain desiccant
oxygen sensors will not be able to precisely determine the argon
content. It may be possible if no desiccant is used or if the desiccant is
preconditioned by saturation in argon prior to use.
-Long term gas retention testing as proposed at IGMA meetings would not
be possible using anything that depended on a stable ratio of nitrogen
and oxygen.
*****
We were attracted to the MOCON OpTech because it's ability to measure
Oxygen is very very accurate and simple. That is still true, but if the ratio of
oxygen to nitrogen is a random variable and unknown there is no way to
calculate the nitrogen accurately. And without knowing the total
contaminate, the mix of oxygen and nitrogen, we cannot accurately
calculate the Argon fill level.