Internal Standard

[shaun78]

I have recently found myself in the unfortunate position of having to dissolve massive amounts of sample into very little solvent for an impurities method that I am developing in order to meet recommended sensitivity requirments of 0.1% nominal.

To make a long story short, I am dissolving 5 grams of product (cream/lotion) into 15 mL of solvent. Not having any 15 mL volumetric flasks on hand I have been adding the placebo product, spiking a known amount via pipet (say 5 mL) and bringing up to volume with an additional 10 mL of solvent delivered by pipette.

All of my recoveries across the assay range (0.1% to 10%) are nearly always 25% to 30% low. I got concerned about this and began looking into the reasons behind consistently low recoveries across the assay range.

Long story short, I made up another sample and pipetted in 15 mL of solvent. I then transfered that into a 25 mL graduated cylinder and just as I began to suspect, I had about 20 mL of liquid.

Clearly I am suffering from volume displacment due to such a large sample size in combination with a small sample volume. There are two ways I can think of to fix this. The first is either to use a correction factor (but then I would need to determine the volume displacment over various weight ranges). The second is to use an internal standard.

Now for the questions after the long history: Would this be an acceptable case in which to use an internal standard? How ugly does this get if I want to say I would be using that same internal standard to quantitate unknown impurities?

I think this is a good case for the use of an internal standard. I am not so sure it is the best idea to quantitate unknowns based off an internal standard. Seems to me like I would need to know a good deal of information about my unknowns before I could really get away with this --- relative responce factor at the least.

Thoughts?

Thanks in advance!

Shaun

[Noser222]

Can't you just scale up a bit and do it all in the 25 mL volumetric flask?

8.3 mL up to 25 mL

[shaun78]

Sadly, no.

The cream takes up so much space that if done in a volumetric flask, there is not enough room to add the cream and solvent to efficiently dissolve the product. I would like to avoid volumetrics all together due to this reason.

Furthermore, due to product cost, I know it is going to be frowned on if I write a procedure asking the analyst to take 8.3 grams to make one sample. Not that 5 grams is much better, but I was barely able to get people to buy off on 5 grams.

With that in mind, 5 grams into a 15 mL flask is going to be nearly impossible. I actually went the other way trying to get 3.3 into a 10 mL flask as I don't have the 15 mL flasks. It did go well.

[Consumer Products Guy]

Can you use a 25 ml graduated cylinder with ground glass top, and go to the 15 ml mark, or even to the 20 ml mark? That might be better than adding 15 ml solvent and ending up with "who knows how many" ml. In one of our assays we do use a graduated cylinder, because we needed its wide neck to get sufficient sample in.

[Uwe Neue]

I like the proposal from consumer products guy. However, there is an alternative that should not be too bad. You can always make a molal solution, by weighing the amount of solvent that you add to the cream sample. Then you need to translate the molal solution to a molar solution. In order to do that, you need to know the density of your final mix of cream and solvent. This can be calibrated, but it may not even be necessary to do that, if the density of the cream and the solvent are identical, or close to identical. Then, in the future, you always make your sample by weight, and refer to a predetermined table to calculate the molarity of the solution.

[HW Mueller]

What is this? 3.3g in a 10mL volumetric goes well, but 5g in the 15mL vessel, which you don´t have, doesn´t go??? Also, you say you got ~ a 20mL solution when adding 15mL solvent to 5mL cream, thus, the suggestion to use a 20mL flask seems a good alternative.

[shaun78]

My mistake. I seem to have left out a crucial "not". 3.3 grams into 10 mL did not go well.

Thank you Consumer Products Guy and Uwe Neue. Those are both interesting suggestions. Both of them seem very parctical and relatively easy to accompligh. I take it that the readings off a graduated cylinder are accurate enough? I would have to demonstrate accuracy of impurities by spike recovery, which our protocols state must be within +/- 10% for impurities.

We shall see how it goes.

Thanks again for the suggestions!

Shaun

[Peter Apps]

As long as the solvent is not very volatile you can do the whole thing by weighing.

Dollop about 5 ml of cream into a 20 ml vial. Weigh it. Squirt about 10 ml of solvent in on top, weigh again. Calculate how much the cream was diluted (in mass terms). Analyse the solution and correct the result by the dilution.

To please your penny pinching masters (is this stuff a suspension of gold dust or something ) you can scale this down by a factor of ten and still have more resolution from a 4 figure balance than you can ever get from volumetric glassware. And vials are cheaper than volumetric flasks.

Peter

[Bruce Hamilton]

I'm with Peter/Uwe on this, I would weigh both, using either a glass vial or ( depending on sample and solvent ) a disposable 16 ml graduated polypropylene centrifuge tube.

They are about $100/1000 ( universally used in biochem/medical labs ) and are wide mouth, if you really want a wider mouth use the 50 ml version conical versions.

I assume you have clear solutions, if not the tubes can be centrifuged to provide a suitable supernatant for injection.

The advantage is these containers can easily be weighed on a 3 figure top-pan balance.

I'd also use a variable-delivery dispenser bottle for the solvent. That means your dispensed solvent volume can easily be matched to sample weight if you want to maintain a constant ratio, especially if the cream is difficult to accurately dispense.

If this was for GC, I would use an internal standard, but I assume even your high dissolved solids-content solutions are giving good repeatability for LC, so an internal standard is an unnecessary complication.

Please keep having fun,

Bruce Hamilton

[shaun78]

Ended up going the weighing route due to the fact that my recoveries were much better using this approach.

Thanks everyone for the input!

[Peter Apps]

Thanks for the feedback.

Peter