GC standard addition questions

[RhysB]

Hi Guys,

I've been looking at standard addition methods for a while now but have not had the time/courage to try one. However I have been forced into it due to problems beyond my control. As such I though I would post what I've done to see if I've done things right, or if I've made some rookie mistake that i can't see.

The samples are water soluble products that are testet for ethanol content (5000 ug/g limit or 0.5%). 1g is disolved in 10 mL distilled water (volumetric flask). I then pipetted 1000 uL into 3 sample vials. To the first I added 500 uL of distilled water. To the second I added 250uL of a standard that I made up that corresponded to 1250 ug/g ethanol and 250 uL distilled water. The third sample recieved 500 uL of the standard which corresponded to 2500 ug/g ethanol. I then injected the 3 samples. The R2 was 0.996 and the final concentration (thanks Excel) was 4260 ug/g. This was calculated adjusting for the dilution of standards and also the actual mass of product weighed in. This was reasonable (!) close to the value that had ben measured by the old methoid (which we are trying to superceed) of 4600 ug/g

Now appart from any howlers that others can spot, I have some questions regarding what I've done. The main one is 'How does one choose the concentration of standard to add to the samples'? I repeated this instead adding 2500 ug/g and 5000 ug/g using the same method and proceedure and got an answer of 4800 ug/g ethanol. In fact thats the only question I can think of at the moment. I'm just trying to learn more about how to use this technique as I think it can solve a few other problems that we have (both GC and HPLC methods)

All comments welcome. Some may be more welcome than others

[chromatographer1]

RhysB

0.996 is a bit low in my comfort range. I would prefer at least 0.998 and I always tried to get 0.999.

But I may be too much of a perfectionist as injecting a non-volatile sample solution can affect linearity of the ethanol response. I always tried to inject 0.5 microliter with a 0.5 microliter of water behind it in the syringe slowly, 1 to 3 seconds. This I found improved my linearity.

Be very careful in preparing your standards. I would calculate the stds as wt of ethanol per volume, directly weighing the ethanol (100% or 95% grade?) into volumetric flasks and filling to the mark.

Rather than making two additions, one being water, just make one addition of the different concentrations. Pipettes have an error. It is always more accurate to determine the density of your solution (1g per 10mL) and then weighing the transfer into a vial instead of assuming a perfect volumetric transfer. These subtle 'refinements' will improve your linearity as well.

I always tried to add 25% of the limit and 50% of the limit as std additions. This keeps the measurement closer to the limit. For example, as a bad practice, adding 5% and 10% alcohol when the limit is 0.5% makes no sense. Likewise, adding 0.05% and 0.1% also makes no sense.

To sum up: minimize the use of pipettes. Use the balance to determine weights in solution of the sampled portions. Your method is good, but I think it can be improved.

best wishes,

Rod

[Peter Apps]

And just to add to what Rod advised - when doing additions by weight you can use an autopipette to add a volume that corresponds to your target addition, then record the actual weight added, and put that into the Excel spreadsheet that does your calibration. You then get the best of both worlds - speed from the autopipette and precision (which should translate into accuracy) from the weighing.

Please let us know how it goes.

Peter

[RhysB]

Thanks very much for the critique. Its always a worry with adapting a methodthat you are not doing something quite right or have missed something fundamental entirely. I do weight in the ethanol but had not considered weighing in the sample and standards to the GC vials. Oh, and I missreported the R2 values, so got a 0.9996 and 0.9998 for a second test. that should be good enough for QA.

What I do really like with this technique is that it has an inbuilt system check. I've been having problems with a random reproducibility error that I'm slowly working through but this gives me confidence in the result, and iff the error pops up I can see it in the graph.

Thanks again for your help.

[chromatographer1]

I am happy to hear the revised report of your experiment.

It sounds like you have a solid method. The self validating nature of the std addition is a comfort for those whose jobs and reputations are at stake.

rest well,

Rod

[krickos]

I agree.

One last caution as your limit indicates working with pharmaceuticals. If you do and plan to continue using excel for release testing instead for a validated software, I strongly suggest validating that excel sheet and locking formula fields etc and document the validation. If not you are bound to get hit on an audit, use of unvalidated excel sheets for release testing is like a homerun for FDA and others.

[RhysB]

Well, I'm happy to report that it all seems to have gone swimingly. The QC technician who did the analysis got R2 of 0.9999 and 1, so she was stoked, and the sample preparation is easier than the previous method. I did have a bit of trouble with the sample concentration calculations (in my defence it was Friday afternoon). Just to clarify a few things. I need to take into account that I'm diluting the samples by 1/3, is there anything else with the additions (apart from initial sample mass of course). The QA chap was doing some calculations to take into account the dilution of the standard addition solvents themselves as well, but I'm not sure if this is necessary.

I now have to come up with some justification to choose standard addition over an external standard method. Is there any sort of text or paper that goes through the pros and cons of both methods? Rod, I've read through all you posts here on the subject, but I'm being asked for something thats been through a pear review process (I think, it had been a long week). Does the FDA have any official or recomended views?

If I can sort this out, then there is some interest to convert most of our final sample residual solvent testing to this method. It will just mean alot of work for me in the new year (at least I'm working I guess).

Thanks again for all your help. Having an online resource like this is like gold, only more useful.

[chromatographer1]

I am unaware of any US Governmental preference to std addition over ext std analytical methods. They tend NOT to dictate and allow the applicant to determine the better method and justify it for their approval. Now this is a good thing.

Consider a few common sense comments:

std addition offers multiple reiterations, at least triplicate, of an examination of a sample. Of course, using different sample solutions - samplings, improve the ruggedness of the test over multiple samplings of a single sample solution.

Doing so offers a demonstration of the homogeneity of the product which increases confidence in the results of your test, demonstrates the reproducibility of the equipment being used and the competence of the technician performing the test.

A 'single rat' test while it may be accurate, also provides a measure of doubt to the accuracy of the test. Was the sample properly prepared? Did the equipment measure accurately? Was there an accidental contamination or mishandling of the single preparation? A reviewer looking at paper results has only what he sees to convince him of the accuracy of the results. Would you believe an eyewitness' account of an event in preference to 3 or 4 eyewitnesses' confirming account of an event?

The more solid foundation we can provide to the results of our use of science strengthens the confidence in the publication of our results and provides a greater margin of safety in the products we produce.

Be glad you are working. I have been unemployed for more than three years.

best wishes,

Rod

[Karen01]

Be glad you are working. I have been unemployed for more than three years.

I was out of work for 1.5 years until about a year ago ... If this job goes away I fear what the future would hold... May I ask your age?

- Karen

[chromatographer1]

I am almost 62.

Rod

[Karen01]

I am almost 62.

I am not too far behind you... I hope you are doing OK.

If I was still out of work I would closing in on 3 years and would be in a very bad financial situation! Getting a job when one is in one's mid 50's and beyond is not easy.

- Karen