[sas1995]

Hello, I am very new to chromatography so I am still learning how to obtain data. Our GC was bought on a budget so no training. I am wondering if a calibration curve using liquid standards can be used to determine gas sample concentrations.

Thanks!

[marquis]

Normally, just injecting the materials gives way too strong a peak.  The materials need to be diluted.  The best way to approach this is to look at the samples or materials you will be testing.  They will require standards.  If your calibration department requires nothing else, start with the standards required for your test methods.  Do you know what columns will be needed?  By the way, what columns will you be using and what are your detectors and injectors?  Thanks.

[rjb]

Hello, I am very new to chromatography so I am still learning how to obtain data. Our GC was bought on a budget so no training. I am wondering if a calibration curve using liquid standards can be used to determine gas sample concentrations.

It's wonderful when companies make the decision to cheap out by deciding to forego the training isn't it?! I have been in a similar position, but despite all the stress that is no-doubt coming your way, you're going to learn a lot very quickly - which is good - but mostly out of necessity!

The answer to your question really depends upon what you're trying to achieve, the nature of your sample, what level of accuracy you require and whether your sample is truly a 'gas' phase sample, or whether you're attempting to run headspace samples and want to know the amount of analyte in a liquid or solid sample.

In an ideal world, X amount of injected analyte should yield Y response from whatever detector your GC uses and this should be reproducible. Moreover, as the amount of analyte X is changed, the response Y should also change and should (hopefully) do so linearly allowing you to produce a calibration 'curve'. In an ideal world, whether your analyte is a 'gas', a liquid or dissolved in a solvent should make negligible difference under the proviso that the same amount of analyte is transferred onto the column and reaches the detector.

This is however where our problems begin... Ignoring the (probably very slight) differences between liquid and gas samples when it comes to retention times (no big deal) due to differences in residence times, there's this nasty little thing called the matrix effect which can cause us all sorts of problems, especially when analyte concentrations are low. When you inject your analyte dissolved in a nice clean solvent (as was implied), active sites (silanols and metals) within the liner (and possibly elsewhere in the system) tend to bind up (or degrade) some of the analyte molecules and these never make it to the column. This means that your calibration curves are based upon less analyte reaching the detector than was intended. This is not a problem under the proviso that the same process happens to the same extent with your actual samples. However, in the real world, samples contain other stuff (the matrix) which might interact preferentially with active sites, meaning that more of your analyte molecules (which in your standards were otherwise bound or degraded) make it to the detector. This could be a considerable source of inaccuracy, but depends very much upon concentration, the nature of the sample matrix, inlet conditions and perhaps even inlet residence time (which would differ slightly between liquid and gas samples). For this reason, calibration conditions and calibration samples should be as close as possible to those used for real samples and it is not unheard of to make use of matrix spiking as part of calibration or using a matrix spiked samples as part of a QC process.

Does all this matter? Well it depends upon what you're doing and why... If this is part of a research project (rather than you providing a proper analytical service, especially one that might have legal ramifications), then it might not be a big deal, you can always discuss it as part of your findings as a source of inaccuracy or attempt to quantify it. Otherwise I think you might like to do a bit of research to find out what others have done before you...

Kind Regards

R

p.s. If you're using headspace and not using 'gas' samples then that's going to need a whole new response!
 

[Babcock_Hall]

It might be worth drawing a distinction between the method of absolute calibration and the method of internal standardization (also called the method of indirect calibration).  The treatment of these subjects in an older book Basic Gas Chromatography by McNair and Bonelli was very clear.  In the latter method the peak area of the analyze over the peak area of the internal standard is plotted on the y-axis.

[sas1995]

Marquis, rjb, and Babcock_Hall thank you so so much for your responses.

Marquis:
Yes, I dilute the compounds in a solvent to make the standard solutions. I get nice peaks doing so. Our GC is an SRI 8610C, with a mid polar capillary column, on-column injection port, and FID/FPD detectors. We are measuring battery electrolyte solvents (ethylene carbonate, dimethyl carbonate, ethyl methyl carbonate, etc.) and their decomposition products to quantify emissions. So, the carbonate solvents are liquid or solids under ambient conditions, which is why I having been preparing liquid samples, but some are also volatile and will produce gases during our our processing with them, which we would like to quantify. The decomposition products are mostly all toxic gases - which is its own issue since we do not have an MS detector and you can't purchase these products to make standards.

Rjb:
It certainly is not ideal! We are a small start-up so money is scarce. I am definitely learning a lot - I did some GC in my undergraduate studies, so things come back to me also here and there. Thank you for the encouragement. Eventually, we will be sampling emissions out of our pilot process, which is what I am trying to mimic right now. Our injection method is manual (another cheap out), so what I am doing right now is using temperature conditions similar to our pilot process (which may decompose and/or volatilize the analytes), and sampling the headspace. Is this even a proper process? Really, I am just trying everything to learn more and obtain responses to present species. I have had no such luck with this method, but I had broken our gas tight syringe so I was using some cheaper ones. I ordered some more gas tight syringes, so I think we will get better results this way but I will just have to wait and see. The measurements don't need to be super accurate, since these numbers don't need to be reported (it is for our own understanding). I will also be taking some liquid samples, which I am not concerned about determining results.
Would you mind explaining how you know when you are experiencing a matrix effect?

Babcock_Hall:
Thank you for the book suggestion, I will definitely get this and try comparing these methods!

[Babcock_Hall]

The chief disadvantage of the method of internal standardization is the need to choose a standard that has a different retention time from the analyte(s).  If someone has already published a protocol for a particular analyte, then one advantage of this method is that an error in the injection volume of the analyte mixture doesn't lead to an error in the amount, because the ratio of analyte to standard is constant.

[sas1995]

The chief disadvantage of the method of internal standardization is the need to choose a standard that has a different retention time from the analyte(s).  If someone has already published a protocol for a particular analyte, then one advantage of this method is that an error in the injection volume of the analyte mixture doesn't lead to an error in the amount, because the ratio of analyte to standard is constant.

The closest paper I've found does not used internal standards, but they also use 2D GC, which we do not have those capabilities. I know acetone (my solvent, and the paper's) does not overlap with any of the analytes, so I could potentially change the solvent to something else and use acetone as the internal std? Is that reasonable?

[Babcock_Hall]

That is a possibility, but you might be better off with the method of absolute calibration.  I would defer to to analytical chemists on the forum for guidance in that area.

[rjb]

Our injection method is manual (another cheap out), so what I am doing right now is using temperature conditions similar to our pilot process (which may decompose and/or volatilize the analytes), and sampling the headspace. Is this even a proper process? Really, I am just trying everything to learn more and obtain responses to present species. I have had no such luck with this method, but I had broken our gas tight syringe so I was using some cheaper ones. I ordered some more gas tight syringes, so I think we will get better results this way but I will just have to wait and see. The measurements don't need to be super accurate, since these numbers don't need to be reported (it is for our own understanding). I will also be taking some liquid samples, which I am not concerned about determining results.
Would you mind explaining how you know when you are experiencing a matrix effect?

Wow! Its sounds like you've really drawn the short straw here... You have my sympathies and my respect for giving it a damn good go under difficult circumstances!

From what I gather, your pilot process possibly produces some volatile analytes (are you able to share what these might be at all?) which you would like to measure. You've used a method - which sounds very much to me to be a perfectly legitimate form of static headspace sampling - which involves (in your case) withdrawing a syringe worth of headspace into a gas syringe and injecting it into your GC (FID, MS, ECD?). So far you haven't been able to spot any analyte in your headspace samples...

As a start, to give yourself a bit of background knowledge you might like to consider this link as a fair start... I recommend the whole site (LCGC) generally: https://www.chromatographyonline.com/view/static-headspace-sampling

There could be myriad reasons for your results or lack thereof.

It might be that your analytes are not there or have decomposed - Clearly something that you have considered as a possibility. Worth further investigation.

Sample volume/method sensitivity might be another possibility - How much headspace are you recovering? My students do a practical looking at headspace analysis of accelerants (normally a simple mixture of neat Octane and Ethanol seeded in quite high volume onto some burnt debris) and with our old teaching GC-FIDs they sometimes struggle getting decent results even when injecting 1ml of headspace! Sometimes this is down to gas syringes and needles (which love to irreversably ingest septum and block up), but it might also be that FID (if that's what you're using?) doesn't have the sensitivity that you need!

Another factor (assuming that your reaction is liquid phase) is that your analytes are very happy dissolved in your solvent and don't want to partition into headspace. You can change the partition coefficient with heat and you could also try salting-out to force more of the analytes to enter headspace and then sample when at their new more headspace favourable equilibrium. That said, switching to dynamic headspace methods or purge and trap (which are not cheap so are probably not feasible) might be your best bet... With these techniques, if the analyte is there, it's going to come out whether it likes it or not! Failing this, have you considered SPME?

In terms of the matrix effect question, you might find the link below to be useful...

https://www.e-b-f.eu/wp-content/uploads/2018/06/fw201709-30.-Benno-Ingelse-Matrix-effect.pdf

From what I can see, this is very much LC focused and GC is a little different (as the number of active sites in the liner does change over time as it degrades), however the principle is broadly the same.

One last point - The use of manual injections (even with liquid injections) almost always requires internal standards (as suggested by BH) as the run to run variability would be too high otherwise. Our GC-MS/MSs with auto-samplers are supposed to have volumetric accuracy; but even with these I always run with IS with my liquid samples, partly because the volumetric accuracy thing is only true in the advertisements and partly to account for changes in MS gain or inlet liner changes. There's nothing more frustrating when after 42 days of stability testing, the technician tells you that they had to clean the MS source and that now the analyte peaks have 120% the area as they did the day before. Not a problem if you use an IS!

In terms of IS the following might be of interest: 

https://www.chromatographyonline.com/view/when-should-internal-standard-be-used-0

This might help address whether acetone would be a good call or not... I suspect not!

Kind Regards

R

[Babcock_Hall]

sas1995,

I think that rjb has covered the topic very well.  I also think the book I mentioned is a good source of general information for people with little prior GC experience no matter which method one uses.

[sas1995]

Our injection method is manual (another cheap out), so what I am doing right now is using temperature conditions similar to our pilot process (which may decompose and/or volatilize the analytes), and sampling the headspace. Is this even a proper process? Really, I am just trying everything to learn more and obtain responses to present species. I have had no such luck with this method, but I had broken our gas tight syringe so I was using some cheaper ones. I ordered some more gas tight syringes, so I think we will get better results this way but I will just have to wait and see. The measurements don't need to be super accurate, since these numbers don't need to be reported (it is for our own understanding). I will also be taking some liquid samples, which I am not concerned about determining results.
Would you mind explaining how you know when you are experiencing a matrix effect?

Wow! Its sounds like you've really drawn the short straw here... You have my sympathies and my respect for giving it a damn good go under difficult circumstances!

From what I gather, your pilot process possibly produces some volatile analytes (are you able to share what these might be at all?) which you would like to measure. You've used a method - which sounds very much to me to be a perfectly legitimate form of static headspace sampling - which involves (in your case) withdrawing a syringe worth of headspace into a gas syringe and injecting it into your GC (FID, MS, ECD?). So far you haven't been able to spot any analyte in your headspace samples...

As a start, to give yourself a bit of background knowledge you might like to consider this link as a fair start... I recommend the whole site (LCGC) generally: https://www.chromatographyonline.com/view/static-headspace-sampling

There could be myriad reasons for your results or lack thereof.

It might be that your analytes are not there or have decomposed - Clearly something that you have considered as a possibility. Worth further investigation.

Sample volume/method sensitivity might be another possibility - How much headspace are you recovering? My students do a practical looking at headspace analysis of accelerants (normally a simple mixture of neat Octane and Ethanol seeded in quite high volume onto some burnt debris) and with our old teaching GC-FIDs they sometimes struggle getting decent results even when injecting 1ml of headspace! Sometimes this is down to gas syringes and needles (which love to irreversably ingest septum and block up), but it might also be that FID (if that's what you're using?) doesn't have the sensitivity that you need!

Another factor (assuming that your reaction is liquid phase) is that your analytes are very happy dissolved in your solvent and don't want to partition into headspace. You can change the partition coefficient with heat and you could also try salting-out to force more of the analytes to enter headspace and then sample when at their new more headspace favourable equilibrium. That said, switching to dynamic headspace methods or purge and trap (which are not cheap so are probably not feasible) might be your best bet... With these techniques, if the analyte is there, it's going to come out whether it likes it or not! Failing this, have you considered SPME?

In terms of the matrix effect question, you might find the link below to be useful...

https://www.e-b-f.eu/wp-content/uploads/2018/06/fw201709-30.-Benno-Ingelse-Matrix-effect.pdf

From what I can see, this is very much LC focused and GC is a little different (as the number of active sites in the liner does change over time as it degrades), however the principle is broadly the same.

One last point - The use of manual injections (even with liquid injections) almost always requires internal standards (as suggested by BH) as the run to run variability would be too high otherwise. Our GC-MS/MSs with auto-samplers are supposed to have volumetric accuracy; but even with these I always run with IS with my liquid samples, partly because the volumetric accuracy thing is only true in the advertisements and partly to account for changes in MS gain or inlet liner changes. There's nothing more frustrating when after 42 days of stability testing, the technician tells you that they had to clean the MS source and that now the analyte peaks have 120% the area as they did the day before. Not a problem if you use an IS!

In terms of IS the following might be of interest: 

https://www.chromatographyonline.com/view/when-should-internal-standard-be-used-0

This might help address whether acetone would be a good call or not... I suspect not!

Kind Regards

R

I really appreciate you taking time to respond to me so thoughtfully. Sorry for my late response. I wanted to try some more experiments before getting back to you.

We have a GC-FID/FPD. I am measuring carbonate solvents common to dissolving battery electrolytes (ethylene carbonate, dimethyl carbonate, etc.) and their decomposition products. Common decomposition products are ethylene, acetaldehyde, and some alcohols. Also, CO, CO2, etc. which are obviously not measurable with our detectors.

With our not-broken syringes (whoops), I am able to measure the carbonate in the gas phase (1-10 uL injection volumes). I believe you are correct that the decomposition products I am looking for may not be there, because it didn't decompose (even though I observed a color change?) or they are at very low compositions.

I also tried making a solution of the "decomposed" analyte in case of dissolved gases (yes, the reactions would be liquid phase), and did not observe anything.

I am definitely making some positive progress, and I really appreciate your time again.

[sas1995]

sas1995,

I think that rjb has covered the topic very well.  I also think the book I mentioned is a good source of general information for people with little prior GC experience no matter which method one uses.

Thank you again!

[rjb]

With our not-broken syringes (whoops), I am able to measure the carbonate in the gas phase (1-10 uL injection volumes). I believe you are correct that the decomposition products I am looking for may not be there, because it didn't decompose (even though I observed a color change?) or they are at very low compositions.

I am definitely making some positive progress, and I really appreciate your time again.

You're welcome
Given the colour change mentioned, I would suggest upping your injection volume significantly and giving it on last shot, if only to confirm to yourself that your analyte isn't actually there at a measurable concentration. Whilst 1ul is plenty for liquid injections, headspace requires much larger injection volumes... 1ml is common (I think it is also default for a lot of the automated pressure/loop based systems which the big boys who do this stuff every day typically use), but a quick google shows that manual headspace syringes of 2.5ml and 5ml are available, indicating that some go above and beyond!

If you're getting results for carbonate at 10ul headspace injection volumes, then the carbonate will most likely saturate your detector with a 1ml HS injection, but at least by injecting a reasonable volume, you can say with confidence that you've injected a commonly utilised volume and got nothing rather than having to report no result with 100th or even 1000th the normal volume of sample.

All the Best

R