linearity preparation

[srdales]

I would like to take a survey.

How many individuals are doing individual weigh out for linearity

and how many are doing dilutions from a stock solution?

[unmgvar]

i would answer that by saying that we are using both with a preference for the first way.
yet we have found that for some applications where we hit a very low concentration or other problems of linearity due to standard preparation we go for the second option which gives more robust results then. mind you in that option we always, always have a control standard concentration in order to show that the initial stock solution was correclty prepared. without the control you cannot trust your linear curve

[Consumer Products Guy]

We use the second way. And after comparing results with manual pipetting v. an automated diluter, and finding the diluter provided great linearity (like supplier claimed), now we use the automated diluter. Like all procedures, if you have crummy volumetric techniques, your linearity will be crummy either way.

[Jumpshooter]

We prepare a concentrated stock calibrant e.g. 5ug/mL then do manual serial dilutions at 1:2 of it with our Eppendorf pipetmen to generate our standards. We have found that if you weigh out decreasingly small and smaller amount of reference standard then add diluent, the stochastic effects predominate and reduce the R-square/fit.

[DR]

If I'm checking a system out, I make a stock and dilute away. If I'm checking a method, I use multiple weighings and make no more than 2 serial dilutions from a given weighing.

[Jackus]

When I saw your comments and recommendation I would like to ask you for your opinion of determining linearity using different injection volumes
(for example 1, 5, 10ul of solution A, 5, 10, 20ul of solution B).
Autosampler linearity is tested during PQ and I suppose the instrument is more accurate than pippeting.

[DR]

I would not have much of a problem with it as we do test linearity of our injectors too, and they're very good - but we usually just pick a set injection volume and go with it. I suspect that the main reason is that, from instrument to instrument, things will go nonlinear at slightly different concentrations due to differences in detector properties (more of a concern than having different columns exhibit signs of overloading at different concentrations).

[Bruce Hamilton]

Depends on sample value.

For items that cost a lot ( eg $100/mg ), I usually weigh into one stock solution, and prepare multiple dilutions - up to five. For low cost samples, I tend to use a couple of weighings and a couple of dilutions from each.

In my experience, good practical technique with suitable dilution devices ( volumetric glassware, air-displacement or positive displacement pipettors, glass syringes ( for organic solvents ) ) tend to provide repeatable results.

I would not normally use the HPLC autosampler for linearity, as samples frequently aren't in the initial mobile phase.

Please keep having fun,

Bruce Hamilton

[AdrianF]

I am going to sound heretical here. I think linearity testing is a waste of time! With the proviso that the max absorbance is 0.5AU and the instrument is properly maintained I cannot see the value of testing the obvious.
Does anyone have any examples of non linearity in these circumstances?

[Jackus]

I am going to sound heretical here. I think linearity testing is a waste of time! With the proviso that the max absorbance is 0.5AU and the instrument is properly maintained I cannot see the value of testing the obvious.
Does anyone have any examples of non linearity in these circumstances?

For assay analysis is max. absorbance 0.5AU suitable but when you analyze impurity profile at 200-210nm with LOD 0,05 area %, you must have major peak about 1-1,5 AU then its linearity must be confirmed ( or use diluted reference as second run). Because when is its response nonlinear you get incorrect result. I observed nonlinearity for some compounds. Mainly near 1,5 AU but in one case the compound behave nonlinear even about 0,9 AU (I supposed it was evocated by its fluorescence). Instrumentation Agilent 1100 with VWD detector.

[HW Mueller]

AdrianF, though I have no specific example "up in my sleeve" I can think of all kinds of things that could aggregate, form micelles or just show the normal interactions which cause deviation from ideality. For instance you may have a low absorbing substance which you need to inject at high concentration, chances are that the concentration is so far from ideal that Beer-Lambert doesn´t describe what hits the detetor. Linearity is also a matter of ideality of the solution, not just detctor physics.

[Consumer Products Guy]

AdrianF - I'm not going to disagree with you, but with using automated diluters and automated samplers, this doesn't require much operator time these days, and the chart looks good.

[Bruce Hamilton]

With the proviso that the max absorbance is 0.5AU and the instrument is properly maintained I cannot see the value of testing the obvious.
Does anyone have any examples of non linearity in these circumstances?

Happens to me a few times every year, although my methods linear ranges can go up to absorbance of 1, with instrument detectors claimed to be linear to >2, eg Agilent DAD and VWD.

The most common examples are methods that do not read on top of a peak ( because of other interferences ), and samples that aren't soluble in the initial mobile phase at autosampler temperature, but ion-pairing methods have sneaked up on me as well - probably something to do with buffer concentrations.

Even with non quantitative methods, I like to have two standard concentrations, just so I know the instrument and methods are behaving.

If you are confident, you could probably initially qualify a method so you don't have to perform a linearity test but, as CPG noted, the graphs do impress clients, and it's simple and cheap insurance.

Bruce Hamilton