Area Reproducibility & Other MISC Horrors...

[july]

I am using a Shimadzu GC-17A/QP-5000. I am running into some error in the measured peak areas - incurred somewhere between sample prep and injection of the sample.

I am trying to analyze different contaminants of indoor air, but we are trying to develop a method that will not require sorbent tubes/sample concentrator. I am having such problems with just elementary stuff..let's see:

I have a 1 L glass gas collection tube/bulb (gct) (a long glass cylinder with stopcocks on both ends and a septum on the long side of "cylinder") at atmospheric pressure. Using a gastight syringe, I inject a volume of liquid toluene or acetone into the gct and allow the sample to vaporize. After 30 minutes, a 100 uL vapor sample is pulled from the gct using a 100 uL gastight syringe and immediately injected into the GC/MS. I have the GCT set up so that at no time am I required to touch the glass - so temperature fluctuations of the sample caused by handling should not be a problem. The septum is changed after every 50 injections.

We would ideally like to use a 10 uL injection size, but the peak areas of what we're looking for are not reproducible with such a small injection volume. We have increased the sample size to 100 uL, and this reduces the percent error - but it's the same variation in areas as when the 10 uL syringe was used -the percent error is smaller though because the peak is now larger with the larger volume injection.

For toluene, I trace out 91 amu and integrate the peak to find the area due to toluene only. We thought of using argon as a standard to compare with...so I traced out 40 amu in the chromatogram but the area of the argon peak fluctuates also - but differently than the toluene peak. I have also compared argon when injecting acetone vapor. Neither of the three fluctuate in exactly the same manner.

In the beginning we thought the gct was leaking, but we have ruled out a leak after much testing. The area values will fluctuate both up and down for consecutive runs.

Someone suggested that the VOCs we were injecting were more dense than air, and were "settling" so the mixture was no homogenous - and thus we were not pulling a representative sample into the syringe. We tried shaking/turning/flipping the cylinder at different rates to prevent this "settling" - but we saw no difference in the amount of fluctuation in the area measurements. We are seeing the same range of areas - just such a HORRIBLE range.

I can make a calibration curve that looks awesome with an r^2 value of 0.97-0.99. But then if I go back the next day and run a sample within the concentration range of the calibration curve - the area is "wrong" according to the calibration curve. I'm jsut not sure why on one day the area is one thing...and the next it's totally different. I measured the temperature in the room - and it's staying between 70 and 80F. I'm not sure if that could be the source of the problem or not.

Any suggestions on any improvements? Thanks in advance!

[JI2002]

How much toluene or acetone do you inject into the 1L glass bulb?

As far as I know, the reporting limits for indoor air are very low, unless you use some kind of concentration technique, you are not able to meet the requirements. If you directly inject 10 ul or 100 ul of the standard from glass bulb and are able to see the peaks in GC/MS, I guess you are making a very concentrated standard. In order to make a homogenous standard, the concentration needs to be low. Also remember to put the glass bulb into a GC oven @ about 60C after adding toluene and acetone. (put syringe in the GC oven also). If you you use internal standard, toluene-D8 should be a good candidate.

[july]

I am injecting 1 to 10 uL of liquid into the 1L gct, so 1-10 ppm samples.

Our plan was to make a sample of known concentration in the gct first. We then would attach the gct to this small test tube shaped vessel with a sidearm stopcock that has been submerged in liquid nitrogen for a certain amount of time and open the tubes to allow the sample from the large gct to be transferred to the small tube with sidearm - which after a time would be allowed to return to room temp before withdrawing sample for injection.

BUT, I am still having problems with the peak areas being reproducible - so I haven't even had a chance to worry about "seeing" less concentrated species....but hopefully we will eventually be headed in that direction.

You said to put the sample and syringe in a GC oven. This is a dumb question...but just a normal oven at 60C would serve the purpose, correct?

Thanks

[JI2002]

Any oven will do as long as the temperature is accurate.

1 ul of neat toluene into 1 L gct will creat a concentration of about 1000 ug/L. That's about 300 ppmv. At high concentration, the molecules will stick to the wall at room temperature. The idea of putting gct in a oven at higher temperature is to force the molecules into vapor phase and make standard more homogenous.

You can try different temperatures to see which works the best, 60C should be a good starting point. You may need to make fresh standard everyday.

[july]

Could you explain the calculation of 300 ppmv please? I'm not sure I follow. Thanks for your patience.

[Peter Apps]

Hi July

When you say that repeatability is poor, exactly how poor in terms of discrepancies between replicates or rsds ?

Part of your problem is the fluctuating temperature of the room and the sample bulb (and the syringe). If you want really repeatable results you have to thermostat everything to less than 1 C variation.

You probably also have incosistent split ratios, or if you are injecting splitless, some eratic losses of sample to the inlet septum purge, caused by pressure fluctuations as you inject.

If I understand your long term plan correctly you are planning to concentrate the organic volatiles by cold trapping. with real samples you will have the problem that water vapour will aslo be trapped, and you end up with an aqueous sample that cannot be injected into a GC without causing all sorts of trouble.

Good luck

Peter

[MikeD]

July

You asked about the ppm calculation. You may have realised by now that your conversion is out by a factor of more than 100x

The ideal molar volume of 92 grams toluene vapor is 24.5 L at 25oC (3.76g/L)
Therefore 1 vol-ppm = 3.76 ug/L
You are introducing 1-10 uL (870 - 8,700 ug) into 1 L air.
Therefore 1 uL toluene/L air = 876/3.76 ppm = 230 ppm (at 25C).

The suggestion that wall adsorption is to blame is probably correct. The further suggestion that warming the flask and the syringe will reduce wall adsorption is also correct. However, it would be very unusual for toluene in indoor air (excluding workplaces) to exceed 1 ppm and typically it's 0.002-0.02 ppm. That's why pre-concentrating sorbent traps are used and that's why ISO went to some trouble drafting such test methods.

[july]

I'm afraid my questions are getting dumber and dumber...but this "vol-ppm" is it the same calculation as ppm v/v or ppmv? To calculate the ppmv would be the volume of toluene divided by the volume of air sample, correct? I am not sure how these calculations differ.

[MikeD]

ppmv, vol-ppm ppm v/v etc are all the same. Inserting the v is an attempt to distinguish a pure gas/gas or vapor/gas concentration unit from other perhaps more confusing units of ppm referring to solid/solid or liquid/liquid. It never refers to liquid/gas as I think you might have done with 1 uL liquid in 1 L of air. I didn't realise that at first.

If you can possibly avoid using ppm in any form please do so. The only reason that air concentration units of ppm or ppb(illion) keep turning up is that sometimes literature sources or regulatory authorities quote them without the conversion. You calibrate most measurement systems in mass units so there should rarely be any reason to involve ppmv. Mass/unit volume at fixed ppmv depends on temperature. Another reason not to use ppmv.

What I think is confusing for those who don't do these unit conversions every day is the first step of the idealised vapour volume that results from the evaporation of a liquid, depending on molecular weight. The volume of the liquid is irrelevant, other than to estimate the mass. Then again if you calibrated the liquid syringe gravimetrically you would eliminate the liquid volume even there and you would just have the dilution flask volume to measure.

[july]

You are amazing! Thanks for your help.

[july]

That's why pre-concentrating sorbent traps are used and that's why ISO went to some trouble drafting such test methods.

Unfortunately, the chemist I work with has read something about the sorbent tubes creating problems and would like to avoid using them if at all possible. It may just be a necessary evil. Besides, we're too poor to buy the ISO docs !

[july]

When you say that repeatability is poor, exactly how poor in terms of discrepancies between replicates or rsds ?

The area at its worst varies plus or minus 10^6 when the peak area is around 10^7.

You probably also have incosistent split ratios, or if you are injecting splitless, some eratic losses of sample to the inlet septum purge, caused by pressure fluctuations as you inject.

The split ratios for each injection are the same. But as for the pressure fluctuations as the sample is injected - is there way to prevent this from happening - possibly using an autoinjector or is there a better way?

Thanks again for your patience.

[Bruce Hamilton]

Have you thought about using a contract laboratory?. If you're too poor to purchase the admittedly exorbitantly-priced ISO standards, what's the point in generating numbers that could also be found by a random search of a local telephone directory?.

Whilst I admire your enthusiasm, I'd be really concrened about the validity of any numbers generated;-
- you don't have good standards ( may be better to purchase a prepared standard from Scotty or similar ).
- your sampling technique will be very variable, as Peter has pointed out, water condensation will be a problem. At best you would have to use some sort of sorbent, either at ambient or cryogenic.
- you can't accurately calculate the concentrations ( internal std may help ).
- you don't appear to have even consulted any of the available information on trace ambient air analysis of low molecular weight aromatics.

In reality, using a hammer to drive a screw is an enormous waste fo energy to generate something that's still not fit for purpose.

I'd be asking the chemist you work with if he has a clue at all.

Bruce Hamilton

[july]

Have you thought about using a contract laboratory?. If you're too poor to purchase the admittedly exorbitantly-priced ISO standards, what's the point in generating numbers that could also be found by a random search of a local telephone directory?.

Whilst I admire your enthusiasm, I'd be really concrened about the validity of any numbers generated;-
- you don't have good standards ( may be better to purchase a prepared standard from Scotty or similar ).
- your sampling technique will be very variable, as Peter has pointed out, water condensation will be a problem. At best you would have to use some sort of sorbent, either at ambient or cryogenic.
- you can't accurately calculate the concentrations ( internal std may help ).
- you don't appear to have even consulted any of the available information on trace ambient air analysis of low molecular weight aromatics.

In reality, using a hammer to drive a screw is an enormous waste fo energy to generate something that's still not fit for purpose.

I'd be asking the chemist you work with if he has a clue at all.

Bruce Hamilton

First of all, Bruce, I'd like to thank you for your kindness - such kind words.

Secondly, I exaggerate our poorness. Unfortunately, I am the low lady on the totem pole, and the decisions to buy ISO methods or use a contract laboratory are not in my hands. I have suggested the use of sorbent; however, my idea was knocked down.

Thanks for your concern.

[Bruce Hamilton]

First of all, Bruce, I'd like to thank you for your kindness - such kind words.

Secondly, I exaggerate our poorness. Unfortunately, I am the low lady on the totem pole, and the decisions to buy ISO methods or use a contract laboratory are not in my hands. I have suggested the use of sorbent; however, my idea was knocked down.

It is possible to measure ambient concentrations of benzene and toluene without using sorbert traps, but I've only experience with trace levels of low molecular weight alkanes and aromatics in volcanic gases.

If you choose not to use cheap sorbent traps, there are expensive specially-treated sampling cannisters used for trace volatile organics.

I'd highly recommend at least reviewing some of the detail in the EPA volatile organics in ambient air procedures, which are freely available.

Perhaps starting with the following:-
http://www.epa.gov/ttnamti1/files/ambie ... o-14ar.pdf

http://www.scottgas.com/
I didn't check the detail of the TO-14 aromatic mixture ( $1300 price scared me ) but there are also Scotty transportable ppm standards for individual aromatics in air at $74 each etc.

Because these tests are routine in environmental contract labs, they have appropriate gas sampling equipment on their GCs, and the price should be reasonable.

Bruce Hamilton