Std Addition for headspace analysis and other techniques

[chromatographer1]

Why chose standard addition analysis?

Here are some thoughts concerning why std addition is a good choice for HS analysis.
Please comment if you disagree, and add more reasons if they come to mind if you do.

Standard addition verifies so much in one set of 3 samples.

You prove that, if a contaminant is there, you can measure it accurately.

You prove (as much as you can with only one experiment) that it is reproducible as the linear regression won't be good if there is too much wobble in the numbers.

You also show that a simple human error did not affect the results. Making 3 consistent human random errors has a low probability of ever occurring.

You also prove that a single non-homogeneous sample did not pass or fail a lot of material upon which you are performing the test

You gain a lot of confidence in your manufacturing and in your product with standard addition technique.

best wishes,

Rod

[Karen01]

Why chose standard addition analysis?

Standard addition verifies so much in one set of 3 samples.

It depends on what you are doing, I think.

When in R&D, if the matrix composition can potentially vary with every sample (screening) then you either you have to to do it for every sample or just rely on an external curve (after examining recovery with a spread of matrices during development) and actually look at the chromatography and not just the results (which you should anyway)...

When you have often have 60-100 samples for screening experiments in a day, I think most would agree an external standard curve is the more practical approach. (and yes my lab often gets 60-100 samples a day for headspace to be run by the same method on a single instrument)

You prove (as much as you can with only one experiment) that it is reproducible as the linear regression won't be good if there is too much wobble in the numbers.

As two points define a straight line, 3 points don't give great statistics.


- Karen

[chromatographer1]

Well, Karen, I guess it is just you and me on this topic.

Concerning your comments:

I agree with every point you made in the context you made it.

So

we agree.

But nobody else gave a response.

Very disappointing.

best wishes,

Rod

[Peter Apps]

Hi Rod

I would add that the choice between standard additions to samples, and running a calibration and then samples "as submitted" depends on the number and uniformity of the samples, and the availability of analyte-free matrix in which to make up the calibration series. With large batches of uniform samples in a matrix that is available without analyte the total number of runs might be lower with a separate calibration than if every sample is run three times (unspiked and spiked at two levels).

A third alternative which is not often considered is to repeatedly run from the same vial, plot peak area vs run number, fit the exponential curve and integrate the area under it and then calibrate that by peak area vs analyte quantity from e.g. split injections. This is extremely time consuming of course, but remarkable accurate and repeatable. I have used it when there was very little of a single sample, and no information at all about likely analyte concentrations to guide standard additions.

I notice a trend on the forum for headspace questions to be about residual solvents in drugs - for these the standard addition approach probably is the most robust and reliable.

Peter

[chromatographer1]

Peter,

You made good points (as always).

Thanks for your input.

Rod

[mckrause]

Your comments about std addn are not limited to headspace analysis. It is, IMHO, the only valid analytical method for varying matrices. In Karen's case she's probably (guessing from other posts) getting 60-100 samples of a very reproducible matrix (QC type work). In my case we are dealing with environmental and food pesticide work, and none of the matrices are reproducible. We lean very, very heavily on std addn in all of our techniques - HS, extractable organics, LC, ICP, etc.

I have tried Rod's approach. It will drive you to drink (if you're not already seated next to me!). It's valid, but giminee Christmas, Rod, how do you keep from going nuts doing it?

BTW, one of the major benefits to us in using StdAddn methods is that you get a relatively accurate assessment of the LOD when you run it.

[Karen01]

It is, IMHO, the only valid analytical method for varying matrices. In Karen's case she's probably (guessing from other posts) getting 60-100 samples of a very reproducible matrix (QC type work).

No it's not QC type work at all ... But it's supporting process development not product release and we can get samples with 10 different matrices on the same day and we don't get told the details of the surveys/time courses being run.

The composition of the matrices can vary a good bit, but every time I've checked recovery with spikes, it has been good enough for what we need.

- Karen

[chromatographer1]

Karen's work is different from analyzing drug samples. It is closer to routine environmental testing.

In both situations, if a determined number if off by 10 or 20% no one gets upset.

If a critical number is required, then std addition can be done on a single environmental sample for precise evaluation, although with such samples will often vary considerably, even from the same site.

In my limited experience with drugs, I have found that the matrix can affect the recovery of an analyte such as a residual solvent, but the difference both as an increase or decrease is usually not more than 20% either way.

If an approximation is needed quickly, then external std or even internal std would be a better technique in order to improve productivity.

I DO BELIEVE in using the best tool for the job at hand. I hope to inform and advise chemists to choose wisely and know and understand WHY they make the choices they prefer.

Thanks to everyone who contributed to this discussion.

Rod

[Karen01]

Karen's work is different from analyzing drug samples.

Though I USED to do/manage analytical development and support for drug development for 14 years, before getting laid off.

In that case for those products for residual solvents by headspace GC, recovery from spiking experiments showed that external standards were fine.

- Karen

[chromatographer1]

For many of the drug samples I tested showed small differences in recovery from one another, often only a few to several percent. Even 20% was not the norm. But it is the exception that is the problem in a tightly government regulated environment.

My bosses were insistent about having data they could defend in meetings with higher-ups and with representatives from the USP and other officials from suppliers. Rather than performing two injections, a sample and an external std, they much preferred doing three std additions. Arguments over data would take longer than performing the third injection.

But often the determining factor is the company policy. Even bosses make mistakes and it is the poor lab worker who gets laid off.

Rod