Performance checks GC-FID method

[Julieke]

Hello!

I have recently joined the forum and am still trying to find my way around it so apologies if this has been the subject of a previous discussion.

I work in the lab of a spirit producer. As part of shipment, we provide the results of GC analysis to the customer, showing the composition of our product. As far as we know, there is no regulation dictating how we should ensure that our instrument is performing and the customer is not asking for any details-we usually make sure that our analysis falls within internal specs.

Our method uses an internal standard and standard curves were obtained a while back for each of our analytes. We do not have any formal internal testing in place but are members of an external testing group, which sends us a sample of a certified amount every 3 month (our results usually compare well with those of other labs). I would like to set up a procedure on a more regular basis to evaluate the reliability of our instrument/method but also look at RSDs on a more regular basis (just in case someone comes asking for them!).

What would be good practice to evaluate the reliability of results on a regular basis? Would be keen to see what other labs are doing.

Thank you!

Julieke

[karbonat]

Dear Julieke,
I would like to share my ideas about your situation as our work is also connected with spirits analysis. Yes, according to EC 2870/2000 the determination of volatiles in spirit drinks is made by GC with the usage of Internal Standard (IS) method. But apart from the traditional IS method there is novel one which is called “Ethanol as IS” and it uses main volatile component (ethanol) for the determination of the same RRF factors. You can find more information in this paper dx.doi.org/10.1021/jf3044956 and on the youtube channel https://www.youtube.com/channel/UCXgL2c ... lW1oxOGqtQ .
Now about your problem. It should be noticed that quality assurance and control procedure are required in item 10 of EC 2870/2000. Standard solution must be used in order to be convinced in the correctness of data obtained in your laboratory. You may create a standard solution from any bought/manufactured spirit, for instance whiskey and use it in laboratories of your testing group. A usual bottle volume is about 1L, it means that you can fill 500 (1000/2) microvials with this spirit and use them as a standard sample. And now what is to be done:

1. You are to compare the obtained results of the measurements of this spirit among testing laboratories and make corresponding conclusions (about any extraordinary laboratory)
2. You are to compare the obtained results of the measurements (at a long period of time, e.g. to make measurements each week for few months) of this spirit within your own laboratory without making any recalibration of your GC instrument in order to find out that RRF coefficients have very similar values.
3. The reliability of your instrument can be confirmed by the calculation of RSD values. It really shows the robustness of method and instrument you use. RSD can be easily calculated from a row of parallel (consequent) measurements (2-4 for instance) according to common formulas.
4. Of course your “standard sample” (a spirit) must be stored correctly (cool and dark place, inert atmosphere is preferred) if you are going to use it at long period of time.

You may examine and test new analytical method presented above. It really makes the analysis much cheaper and simpler, and metrological characteristics are more stable than in traditional IS method. You are welcome to ask me any additional questions (karbonat7@gmail.com).

[Peter Apps]

Hello!

I have recently joined the forum and am still trying to find my way around it so apologies if this has been the subject of a previous discussion.

I work in the lab of a spirit producer. As part of shipment, we provide the results of GC analysis to the customer, showing the composition of our product. As far as we know, there is no regulation dictating how we should ensure that our instrument is performing and the customer is not asking for any details-we usually make sure that our analysis falls within internal specs.

Our method uses an internal standard and standard curves were obtained a while back for each of our analytes. We do not have any formal internal testing in place but are members of an external testing group, which sends us a sample of a certified amount every 3 month (our results usually compare well with those of other labs). I would like to set up a procedure on a more regular basis to evaluate the reliability of our instrument/method but also look at RSDs on a more regular basis (just in case someone comes asking for them!).

What would be good practice to evaluate the reliability of results on a regular basis? Would be keen to see what other labs are doing.

Thank you!

Julieke

Alcoholic beverages are among the most strictly controlled commercial materials (mainly because the excise duty and other taxes depend on their alcohol content) and so there must surely be official guidance on quality control.

Something that would be easy to do is to use the leftovers from the interlab tests (since these have a known content) and inject them with each batch of samples. Set a maximum - minimum range for the result.

Are you measuring ethanol content or impurities ?

Peter

[Julieke]

Our GC method is used to determine the amount of low-level alcohols (and esters)-the level of ethanol is measured by a density meter. As Karbonat indicated, directive EC 2870/2000 outlines accepted methods for GC analyses and highlights the need of a QC test. To my understanding, this gives an indication of accepted methods but labs can use alternatives as long as they are scientifically valid (?). I am unsure of how often standards should be re-tested or how often RSD should be checked.

Peter, your suggestion of using the leftovers would make things easier, as this gives us a certified spirit and a range of accepted values for each. We run samples at an inconsistent frequency, sometimes 10 a day but sometimes don’t run any for a few days at a time. Would you run standards every 10 samples (for example) or would you run one every day to check for instrument variations?

Thanks for the help!

[karbonat]

Dear Julieke,

according to item 8.5 of EC 2870/2000 it’s necessary to insert one QC standard every 10 samples in order to ensure analytical stability.

RSD plays one of the final roles in quantitative analysis, acting as a metrological parameter which is usually given in the presentation of measurement results (analysis) and which characterize their reproducibility. So it is not an instrument parameter that should be checked within each N measurements or days. I’m pretty sure that you do duplicate or triplicate GC measurement of each analyzed sample, that’s why you may calculate RSD each time and show the results to the customer, if necessary.

[Julieke]

Thanks for pointing me to this, I had thought it was to be done during linearity testing but had not even considered sample testing!

We don't do any consistent replicates testing on our actual samples. We only do it on our certified samples occasionally to check how reliable our method is.

[Peter Apps]

Are you running a calibration with each set of samples ? If so, this gives you a measure of method and instrument performance on the day you ran them. Do the guidelines require more than this ?. I would not run check standards on days when there were no samples - if time allows you could run a set of five before each sample batch (or interspersed between samples) to give a good estimate of repeatability. Once you have done sets of five a few times you will have a very good picture of how stable your system is within and between days, and you can then reduce the number of check standards you run if the system is stable.

Peter

[Julieke]

We currently don't run a calibration or check repeatability with each set of samples-we check our external testing performance and occasionally do a set of standards. The number of samples we run each day is pretty inconsistent and maybe we wrongly thought that it seemed a bit of an overkill to run standards when you only analyse one sample. Regulations only mention running a number of QC/standards between a series of samples but this does not really work if you run smaller numbers.

I think that we will try what you are suggesting, start with sets of 5 standards every day when we run samples and decrease frequency as we get an idea of stability.

Thanks for the help, it has been very helpful.

Julieke