Internal Standard for Acetonitrile detection

[miultimodia]

Hi,

In our laboratory we are trying to detect acetonitrile and ethanol in a water solution and we were using methanol as internal standard. The calibration curve went great for the ethanol but we are having too much trouble with the acetonitrile repeatability.

I couln't find much information but we are trying to use 2-methoxy ethanol (1) or isobutanol (2) as internal standards, besides methanol for the ethanol detection.

Before we try with these standards I would like you to tell me if you had any experience with the acetonitile detection and if you could suggest any other internal standard.

Thanks a lot!

1. Determination of Acetonitrile in Mixtures by GC, Joshipura (1982)
2. http://www.ineos.com/Global/Nitriles/Pr ... gc-ReB.pdf

[Johnny Rod]

Are you using the method in the second reference, but water instead of acetonitrile as solvent? What size injection are you using 1-2ul? It's not the right column to use for solvents especially with water around, you'd be much better off with a PEG type phase.

Any internal standard will do as long as it is well resolved from the analytes and matrix, and is not an impurity in the materials being analysed.

[miultimodia]

Thanks for the reply.

The sample is a solution of [18F]-Fludeoxyglucose in water and saline solution, with ethanol and acetonitrile as residual solvents. The column is a factorFOUR VF-624ms. The size of the injection is 1ul.

The problem with the acetonitrile appears when we try to make the calibration curve: we have to inject many samples to get a proper regression value. This problems does not appear with the ethanol (five injections, five points, r2=0,9999).

[Consumer Products Guy]

I would like you to tell me if you had any experience with the acetonitile detection and if you could suggest any other internal standard.

Have I had experience? Ha!

Anyway, USP 611 details using ACN as the internal standard for ethanol detection using a 624 capillary. While there are some facets of that procedure that I'm not wholly in agreement with, seaparation of ACN from the ethanol is good. Also good is the separation of n-propyl alcohol using those conditions, elutes latest of those three, so that could be used as internal standard for both.

Ethanol and ACN also separate well on PEG column, as do ethanol and n-propyl alcohol, but I can't rmember if I've done the mixture of all three on PEG.

Have fun.

[tom jupille]

I missed something here. If the ethanol curve looks OK but the acetonitrile does not, why would you expect that changing the IS would help? Internal standardization corrects for correlated errors and it would appear that the ACN errors are uncorrelated in this case.

[miultimodia]

I missed something here. If the ethanol curve looks OK but the acetonitrile does not, why would you expect that changing the IS would help? Internal standardization corrects for correlated errors and it would appear that the ACN errors are uncorrelated in this case.

Thanks for all the answers.

Tom, maybe I am wrong, but it seems to me that the methanol internal standard does not reflect the possible "artifacts" that the acetonitrile is suffering, that's why I thought of adding another internal std that would behave closer to the ACN.

Do you think that my problem is with the method I'm using?

[Consumer Products Guy]

The problem with the acetonitrile appears when we try to make the calibration curve: we have to inject many samples to get a proper regression value. This problems does not appear with the ethanol (five injections, five points, r2=0,9999).

What minimum R2 number does one need on a trace level analysis such as this? I'm wondering if you're expecting too much from an impurities level assay.

[tom jupille]

another internal std that would behave closer to the ACN

The catch to that is that you have to make an educated guess about the dominant source(s) of error in the acetonitrile and then find something that would behave the same way. So you have to figure out what's going on: noisy baseline? overloaded tailing peak? getting down near LOQ (as CPG suggested)? variable detector response? . . . [Actually, looking at that list, I'm not sure that an IS would help any of them.]

Ultimately, the only compound that behaves *exactly* like acetonitrile is acetonitrile (there's a reason the MS people use stable-isotope labelled compounds as ISs).

[miultimodia]

What minimum R2 number does one need on a trace level analysis such as this? I'm wondering if you're expecting too much from an impurities level assay.

Finally we were able to get an r2 value of 0,998, and that's great. The problem is that is hard for us to get to that level, we have to repeat many injections to get to this number.

The catch to that is that you have to make an educated guess about the dominant source(s) of error in the acetonitrile and then find something that would behave the same way. So you have to figure out what's going on: noisy baseline? overloaded tailing peak? getting down near LOQ (as CPG suggested)? variable detector response? . . . [Actually, looking at that list, I'm not sure that an IS would help any of them.]

Ultimately, the only compound that behaves *exactly* like acetonitrile is acetonitrile (there's a reason the MS people use stable-isotope labelled compounds as ISs).

The problem is the repeatability of the injections. I would try to find what is the issue with our method, since you think the solution is not to add another internal standard.

Again, thanks a lot for all the answers!

[tom jupille]

The problem is the repeatability of the injections.

If you are using splitless injection, I would be surprised at that, because that's a situation where internal standards *do* help. I think about it this way: if my injection volume varies, all the peaks will be affected the same way, and the errors will cancel out when the area ratios are calculated. If you are splitting, then you may have an issue (perhaps inlet temperature?) that is affecting the acetonitrile differently from the other peaks.

[dpr]

Hi,

Are you using an autosampler?

We use the restek columns not only for product Acetonitrile work (105 meter) but also process streams (30 Meter columns). These are used for low ppm Aceto in samples up to, as listed on web site, impurities in product stream.


An alternative column could be poraplot Q.
We used this to before moving to the restek column to analyse Acrylonitrile product and process stream samples.

Another choice of int std might be acetone. Close elution to Aceto on the columns discussed and readily available. As long as it's not co eluting with another component.

[James_Ball]

Thanks for the reply.

The sample is a solution of [18F]-Fludeoxyglucose in water and saline solution, with ethanol and acetonitrile as residual solvents. The column is a factorFOUR VF-624ms. The size of the injection is 1ul.

The problem with the acetonitrile appears when we try to make the calibration curve: we have to inject many samples to get a proper regression value. This problems does not appear with the ethanol (five injections, five points, r2=0,9999).

Could your problem here with ACN be coming from the saline solution? I know that if you add salt to the point of saturation you will separate ACN from Water when doing extractions. If your salt content in the saline solution is high it could be making the ACN less soluble and therefore less stable in solution. Try with using only water and see if you get better results and that will tell you if the saline is causing the problem.