Find maintenance, best practice, and troubleshooting articles.
Hi, I am developing a GC/FID method for a 6890 that characterizes and quantitates halogenated analytes used in vapor degreasing. My problem is that my results are not precise and I do not have a calculation method to express my results as weight percent (
Just a couple of thoughts. Split injection using volatile solvents like DCM (bp 40C) as mentioned is notoriously unreliable - Effects like small bubbles in the needle and the position of the needle/vaporised cloud inside the liner will cause differences in the actual amount injected or split (your 1µL of solvent will expand to ~450µL of vapour) and can easily give the sort of errors that you are seeing when an External Calibration method is used. If you can find a suitable Internal Standard (IS), I would expect that your accuracies should be within 1% even with relatively large variations in the amount actually injected.
As an experiment, if we took one of the analytes (like butylene oxide) and said it was an IS (obviously it isn't) we can do a ratio calculation like 1% =162.65/352.27 = 0.46172; 5% =665.229/1671.84 = 0.3979; 10% =1135.15/2785.93 = 0.4071; 25% =2772.38/6994.22 = 0.39638; 50% =5100.6/12672.9 = 0.40248; 2nd 25% 2523.58/6455.2 = 0.39094. Let us ignore the first one (1%) because it seems to be an outlier, then if we take the rest we get a mean of 0.4094 and a Standard Deviation of 0.006153 which is a Coefficient of Variation of 0.015 or 1.5%.
Manual inspection of the area counts of your 1% (outlier?) sample for dichloroethylene and butylene oxide (162.6 and 352.3) compared to the 5% sample (352.3 and 1671.8) are roughly 0.244 and 0.217 and compared to the 10% sample are roughly 0.4019 and 0.398 which are very roughly what I would expect - The calculated amounts of 0.000 for these analytes shows that you also seem to have a problem with your integration/calculation parameters.
Because we have not used a constant amount of our "pretend IS" butylene oxide, but varied it according to the amount of our "unknown" analyte dichloroethylene, we would expect the results to be worse than if you had used a constant amount of IS. Normally you would choose an IS to be a material that is generally similar to your unknowns, the peak should be resolved, and have a retention time that is in a generally similar window to the unknowns. Typically the amount of internal standard chosen would give a peak area very roughly equivalent to that of half way up the calibration curve for your analytes.
I would expect that a suitable internal standard (MEK?, fluorobenzene?, cyclohexane??) added to your solutions at roughly the same amount that you used for your 25% standards (the same ~25% for all calibration levels and unknown solutions) should give you a CoV of better than 1%.
Just a couple of thoughts. Split injection using volatile solvents like DCM (bp 40C) as mentioned is notoriously unreliable - Effects like small bubbles in the needle and the position of the needle/vaporised cloud inside the liner will cause differences in the actual amount injected or split (your 1µL of solvent will expand to ~450µL of vapour) and can easily give the sort of errors that you are seeing when an External Calibration method is used. If you can find a suitable Internal Standard (IS), I would expect that your accuracies should be within 1% even with relatively large variations in the amount actually injected.
As an experiment, if we took one of the analytes (like butylene oxide) and said it was an IS (obviously it isn't) we can do a ratio calculation like 1% =162.65/352.27 = 0.46172; 5% =665.229/1671.84 = 0.3979; 10% =1135.15/2785.93 = 0.4071; 25% =2772.38/6994.22 = 0.39638; 50% =5100.6/12672.9 = 0.40248; 2nd 25% 2523.58/6455.2 = 0.39094. Let us ignore the first one (1%) because it seems to be an outlier, then if we take the rest we get a mean of 0.4094 and a Standard Deviation of 0.006153 which is a Coefficient of Variation of 0.015 or 1.5%.
Manual inspection of the area counts of your 1% (outlier?) sample for dichloroethylene and butylene oxide (162.6 and 352.3) compared to the 5% sample (352.3 and 1671.8) are roughly 0.244 and 0.217 and compared to the 10% sample are roughly 0.4019 and 0.398 which are very roughly what I would expect - The calculated amounts of 0.000 for these analytes shows that you also seem to have a problem with your integration/calculation parameters.
Because we have not used a constant amount of our "pretend IS" butylene oxide, but varied it according to the amount of our "unknown" analyte dichloroethylene, we would expect the results to be worse than if you had used a constant amount of IS. Normally you would choose an IS to be a material that is generally similar to your unknowns, the peak should be resolved, and have a retention time that is in a generally similar window to the unknowns. Typically the amount of internal standard chosen would give a peak area very roughly equivalent to that of half way up the calibration curve for your analytes.
I would expect that a suitable internal standard (MEK?, fluorobenzene?, cyclohexane??) added to your solutions at roughly the same amount that you used for your 25% standards (the same ~25% for all calibration levels and unknown solutions) should give you a CoV of better than 1%.
Thanks for your reply Tim. I am curious about your thoughts regarding DCM. If DCM is not a good solvent in this particular case, what would be a suitable solvent? I would generally need a solvent miscible with other low boiling azeotropes. As for using an internal standard, I do not have a problem with it. My original question on this thread however was regarding how I should go about preparing calibration stock, whether for all analytes in question, or a calibration stock for each respective compound in the analysis. Up until now, I was preparing a calibration stock with equal amounts (100g) of each analyte (400g total), but I was informed that the actual amounts for each analyte would not be over 25.0%w/w. This is no good considering that the upper limit of my calibration range must be up 98.0%w/w.
Hi megaflorch, I have read this entire string and hope to add some insight.
1. As you have learned, you cannot make a stock standard near 100% with more than one analyte. When you add components, you dilute the concentration of each analyte. Adding 100ml of any anlayte with a final volume of 1L, regardless of what else you added, that single analyte is only 10%.
2. Why does your upper limit need to be 98%? Is it because some samples will indeed be that high in a single concentration? You mentioned that most samples have 4 analytes. If there are 4 analytes, the individual concentrations cannot add up to more than 100%.
3. So, you could keep 25% as your top cal range, and run your samples at 1x (1ul) and 2x (0.5ul)
4. Or you could inject 2ul of the 25% solution to achieve 50% as the top of your calibration.
5. Or you could go a completely different route and do single calibration - and get 100% linearity with a new approach.
Can I know what components you are calibrating for? I can help you with resolution (get rid of double peaks, with just a few tweaks to your GC Method)
Let me know if you still are looking for solutions for your method development. The goal for you being - simple calibration, no dilutions, no re-runs, and good data.
Thanks for your reply jacobson. I have taken a slightly new approach to this issue, let me know if you think this adequate for what I want.
My new stock standard has four analytes: decafluoropentane, tetrafluoroethyl trifluoroethyl ether, 1,2 trans dichloroethylene, and 1,2 butylene oxide. Since all of my analytes are at 99.9% purity, I translated that value to mean 999000 ug/g for each analyte. So for my stock standard mix, I used an analytical balance read to 0.0000g and weighed out precisely each analyte. this mixture I then diluted 1:2, allowing each analyte to exist in the mixture at a set concentration of approximately 499500 ug/g. I thought this concentration would eliminate my partition problems, as each GC vial holds approximately 2000000 ug per vial. When I ran my calibration, it passed, though I had to eliminate my level 6 standard (99.0%). The chromatography for my runs is very good, I am getting good separation and nice, thin, symmetrical peaks and very little column bleed.
I suppose that I do not have very much experience in working with Chemstation, (B.04.03(16)) my version, in terms of setting up and understanding the processes involved in calculating results. I would like my results to either be able to be read as percent weight w/w, or in ug/g. Is there somewhere in chemstation where I can enter in values or weights where the software will calculate my results with minimal fuss? Any insights or assistance that you can give would be appreciated, especially with working with chemstation.
It looks as though you are on track. As posted above, the way to go to near a 100% concentration of an “unknown analyte” is to dilute your unknown in the same solvent as your standards as suggested, so you might weigh out only 0.100g in your vial and add 1500ul of solvent. DCM is a common solvent but because of its low boiling point is not always ideal for accurate quantitation work unless you use an internal standard - If a bit of solvent evaporates, or bubbles form, it doesn’t matter as your analytes and IS will have higher boiling points and will stay in the same ratio. A suitable solvent with a higher boiling point might be something like chloroform or CCl4 (BP 61 & 77), but obviously (like DCM) they have significant toxicities; you would also need to check that these higher boiling solvents elute away from your analytes.
As a general point DCM is often used for environmental work to extract low levels of pollutants from environmental samples as it will dissolve most organic analytes, but it is almost always used with an IS. Other potential problems with DCM include its reactivity with trace levels of water in hot injectors where it can form HCl which can cause high levels of activity and it will reduce the life of your column; and as mentioned it produces relatively large vapour volumes (~450ul/ul) vs heptane (~200ul/ul); by the way MeOH is worse at ~750ul/ul. A general rule for a longer column life and to keep a relatively low level of activity in your injector is to use a hydrocarbon solvent like heptane, cyclohexane, or toluene, rather than a polar or halogenated solvent; but obviously not everything will have a high enough solubility in apolar hydrocarbon solvents, and they are not really suited for polar stationary phrases like Carbowaxes...
Thank you for the reply Tim. The information you gave regarding DCM is helpful. My reasoning for using DCM is that I wanted to make sure that my analytes were completely dissolved. Also, the analytes that I am testing for are all low-boiling, flourinated, azeotropic solvents used in vapor degreasing. Right now, I am searching for a suitable internal standard. I am having trouble picking a compound that has a retention time outside the range of my analytes: about 1.34-1.90 mins. I have adjusted my inlet programming to account for the problems associated with DCM, like a large split ratio, an inlet liner with taper and glass wool, and I keep a close eye on my split inlet purge filter. With these adjustments, I can expect that my chromatography will achieve good separation and thin, symmetrical peaks. If you have any suggestions for a suitable I.S., I am open. Thanks again.
We are glad this was helpful! We are sorry this was not helpful. If you still need assistance please create a community post or contact support. To help us improve, please provide any additional feedback.
For full details of how we will treat your information, please view our privacy policy.
