Chromatography Forum

So the problem starts, we need a method for 10 residual solvents by GC...

Compound is VERY insoluble in water, soluble in most organics though. Acetic Acid is one of the necessary components to evaluate. Trying most solvents at 100ppm for validation, benzene at 1ppm. Acetic Acid is the only one giving fits. Given the water insolubility of the compound (it doesn't work with most HPLC options), our other options for evaluation are small, we don't routinely do NMR.

Rtx-stabilwax, 30x0.53x1. DMF diluent, direct 180*C splitless injection, 200*C FID detection, tapered glass wool liner, 1uL injection. Went with External Standard because the internal standard seemed to be an issue (maybe jumped off that bandwagon too soon though). The standards look fabulous, all peaks <2% RSD over 6 injections.

Move on to samples and accuracy/recovery: All compounds are recovered at 100 +/- 10% EXCEPT Acetic Acid. Now usually I would assume consumption but it's actually coming out at 200% recovery. There is none in the blank sample injection.

Any ideas as to why it's showing up so high? I was going to try validating at a higher ppm limit, say 500 ppm or so and see if it helps, we have up to the 5000 ICH limit but you know consumers, they want to see none so we need it validated lower if at all possible. Or using an acidified diluent? Injector/detector temps?

Thoughts and suggestions welcome. Thanks!

Hi

Ruling out basic physical properties can save time sometimes. Done some in process testing on drug substances over the years and not all are suiteble for direct injection, they can break down and degradation products interfer with chromatography giving false high values due to either coelution or formation of the solvent you look for.
If you do not have access to it ask for like DSC data and state your issue to the customer, occanssionally just looking at the molecule might give you a hint.

Recovery at 200% ?

I suspect your recovery of std is low, not that your sample is high. You should be doing linearity calibration by std addition.

Acidified dissolution solvent is mandatory. Sulfuric acid is suggested.

Good luck,

Rod

Thank you for the replies!

Here's the plan for now, I'll report on results for the benefit of anyone who might search this out.

1. Acetic Acid as it's own standard, separate from the other 9 solvents, to eliminate any other possible interactions (with and without sulfuric acid modification of the DMF).

2. Acetic Acid at a higher level in the standard, possibly 100 ppm is just outside the linear range (although it did appear linear, still trying this to see).

3. A spike and recovery calc (on an individual sample basis) like the USP uses for Benzene. I've never validated a method for that so I'll be on a learning curve if we end up with that. I think this is what you were kind of getting at, right Rod?


Things I would have LIKED to utilize but probably won't work for us:
1. Would have LIKED to try using the compound in the ESTD standard. However, we don't have clean enough reference or enough of it to do this on a long term basis without qualifying a new purified sample. I think that would really have leveled the playing field.

2. Acetic Acid as a limit test. It's linear, precise and repeatable, but since we can't recover at the right amounts on spike, limit testing at say 500ppm would have been a good answer I think.

You, AnalystX, wrote:

3. A spike and recovery calc (on an individual sample basis) like the USP uses for Benzene. I've never validated a method for that so I'll be on a learning curve if we end up with that. I think this is what you were kind of getting at, right Rod?

My understanding of the present USP procedure is that it is a limit test method. I prefer a std addition method which eliminates the problem of low or high recovery.

But if you take your present sample solution and spike it at three levels of HoAc, adding the same amount of solution and spike solution to your vial you will be able to perform a linear regression and determine the amount of HoAc present in your sample. (as long as the regression is linear. I prefer R squared equal to 0.995 or better)

Search the archives for details. I have helped some on the Forum overcome very difficult issues with difficult samples.

best wishes,

Rod