[WinterNite]

Hi everyone!

I am currently using a GC-MS instrument to help determine the conversion and yield of a catalysis reaction.

In order to do this, I have constructed an internal calibration curve using several prepared samples of a constant concentration of the internal standard, and varied concentrations of the desired product of the catalytic reaction.

In order to determine the yield of the catalytic reactions, I take an aliquot from the reaction mixture, and dilute it with DCM and inject it into the GC-MS. Based on the response and comparison with the internal calibration curve, the yield of the reaction is determined.

However, the issue I face now is that the yield determined via the GC-MS seems to always be significantly lower than the isolated yield of the desired product. I am not that familiar with analytical chemistry instruments, and so I am wondering what the possible problems are resulting in this underestimation?

Cheers

[Babcock_Hall]

Are you using the GC areas (as detected by FID, for example), or are you using some other quantity?  I am not sure why it would matter, but I may be missing something.

[rjb]

Hi WN,

Unless something has gone wrong in your set-up, I suspect this relates to what is often described rather mysteriously as the matrix effect - something well worth getting to grips with!

The matrix (your reaction mixture) is not an inert entity and there will be interactions at various points with the instrument (liner, column etc.) and analyte that will affect overall instrumental response. You can partially correct for this in several ways such as using matrix matched calibration (approximating reaction mixture conditions rather than just diluting your analyte in nice clean solvent) or carrying out standard addition.

You might find this to be of value...

https://www.lgcgroup.com/media/2997/preparation-of-calibration-curves.pdf?ipignore=true

Kind Regards

R

[WinterNite]

Are you using the GC areas (as detected by FID, for example), or are you using some other quantity?  I am not sure why it would matter, but I may be missing something.

Hi there. I am using GC areas to determine the yield. More specifically, the calibration curve I have constructed is based on relative response factors - meaning to say the ratio of the response of the analyte to the response of the internal standard (meaning the area under the peaks in the GC-MS spectrum) is plotted against the ratio of the concentration of the analyte (unknown) to the concentration of the internal standard (known).

[WinterNite]

Hi WN,

Unless something has gone wrong in your set-up, I suspect this relates to what is often described rather mysteriously as the matrix effect - something well worth getting to grips with!

The matrix (your reaction mixture) is not an inert entity and there will be interactions at various points with the instrument (liner, column etc.) and analyte that will affect overall instrumental response. You can partially correct for this in several ways such as using matrix matched calibration (approximating reaction mixture conditions rather than just diluting your analyte in nice clean solvent) or carrying out standard addition.

You might find this to be of value...

https://www.lgcgroup.com/media/2997/preparation-of-calibration-curves.pdf?ipignore=true

Kind Regards

R

Hi R. Thank you for the response. The document you provided seems very useful, and I'll definitely have a closer read.

I'll need to have a think about the matrix matched calibration method, and how much the matrix effect affects my results. At present, what I have been doing is taking a very small aliquot of the reaction mixture, extracting it with water, then diluting the extracted organic phases to standardize the volumes and thus concentrations of the products in each sample. The sample is then used for GC-MS analysis to determine yields. Based on this procedure, I feel that the matrix is, for the most part, largely similar to using a clean solvent used in the construction of the calibration curve.

I imagine that the matrix effect is still present, but it feels unlikely to result in differences of almost 20% in terms of response. But, I am quite unfamiliar with this so if you have any other thoughts, I'd be happy to hear them.

Thank you once again for your input!

[rjb]

Hi WN,

Although matrix effects can be profound and can give rise to results which differ by >20%, I think - given your process - it may well be worth switching attention to the extraction and sample prep steps; making the assumption of course that the 'isolated yield' determined by whatever other methods you used is correct. 

I guess the big question is, how effective is your extraction and sample prep? Liquid-liquid extraction is unlikely to be 100% effective, so I wonder whether the difference could be related to incomplete extraction?

Some people add a form of internal standard (sometimes called a surrogate) before the extraction and sample prep steps and use this sometimes in addition to a conventional internal standard. By using a well-chosen surrogate (ideally but not necessarily a deuterated version of your analyte) which undergoes the same processes as your analyte, inefficiencies in extraction can be determined and accounted for as the concentration of the surrogate prior to extraction and sample prep is known.

Kind Regards

R

[WinterNite]

Hi WN,

Although matrix effects can be profound and can give rise to results which differ by >20%, I think - given your process - it may well be worth switching attention to the extraction and sample prep steps; making the assumption of course that the 'isolated yield' determined by whatever other methods you used is correct. 

I guess the big question is, how effective is your extraction and sample prep? Liquid-liquid extraction is unlikely to be 100% effective, so I wonder whether the difference could be related to incomplete extraction?

Some people add a form of internal standard (sometimes called a surrogate) before the extraction and sample prep steps and use this sometimes in addition to a conventional internal standard. By using a well-chosen surrogate (ideally but not necessarily a deuterated version of your analyte) which undergoes the same processes as your analyte, inefficiencies in extraction can be determined and accounted for as the concentration of the surrogate prior to extraction and sample prep is known.

Kind Regards

R

Hi R

Thank you again for the response. I think the efficiency of the liquid-liquid extraction probably isn't the issue at hand. I say this because the yield of the desired product determined by the GC-MS is lower than the isolated yield determined by the purification process I described. This means that even if the extraction process is not 100% efficient, it shouldn't be the problem at hand. If anything, it only means that the difference between the yield determined via GC-MS analysis and the isolated yield is even larger than I currently have determined.

That said, your information about the surrogate is new to me and very useful still. Thank you for that.

Cheers
WN

[rjb]

Hi WN,

That makes sense... I wasn't aware that your other method used the same extraction process.

As a last ditch attempt, how confident are you that your IS and analyte are both thermally stable and that your GC remains comparatively inert?

In situations when I suspect active site formation affecting IS and maybe analyte (which is a possibility), I normally change the liner and septum and then snip off 10-20cm of column to see if this improves the situation. 

[Babcock_Hall]

I wrote up a document that introduces the isotope dilution method of analysis, which I think was a method to which rjb alluded, on a very basic level.  I am happy to share this with anyone who is interested.