Chromatography Forum

A sample preparation protocol I use for HPLC is rather simple and straight forward with good recovery and is reproducible. However, it involves a dilution step and sometimes this dilution step can be a source of error. The solvent is an alcohol and sometimes can be difficult to pipette accurately.

Regardless I've tried to do some experiments to see if an internal standard can be used to 1- notice when this error occurs and 2- correct for it. Normally the analyates are quantified with an external standard and there is no real reason I need to use internal standard for that sort of quantification.

The internal standard (IS) is added to the extraction solvent in a known concentration. If I compare a set of samples (n=10) where I perform a dilution and don't perform a dilution the %RSD of the IS peak area is 2.4% for diluted samples and for undiluted stock sample 0.9%. This suggests that the dilution step introduces some variation. In larger sample sets I've seen the %RSD go up to about 6% for the IS peak area.

I'm trying to think of best way to develop a correction factor for dilution error. Internal standard in solvent at 100ug/mL concentration. One thing I've tried is every set of samples generate a response curve for the internal standard in that batch of solvent over a range that covers undiluted and diluted samples 10-100ug/mL. Get a linear regression y = xb. If I made a 5 x solution I should have 20ug/mL IS. So I calculate what the area of IS would be if the IS was exactly 20ug/mL and get an expected peak area value. I then divide the expected peak area value with the actual peak area value and obtain a correction factor. I then use this correction factor to correct peak areas of analyates.

Does this make sense? How can I prove whether or not using the internal standard is more accurate then not using it? Is there a simpler way?

Thanks for any help anyone can provide.

If I understand correctly you make up a solution of IS in solvent, and then use that solution to dilute the samples.

Whether this will work or not depends on the ratio of sample to diluent.

e.g. If you dilute 1 part of sample with 100 parts of spiked solvent the concentration of IS in the diluted sample will be nearly independent of diluent volume and sample volume, and so the IS peak size will be constant, and no good for correcting errors.

If you have nearly equal volumes of diluent and sample the IS peak area will be smaller with a higher sample: diluent ratio, while the analyte peak will be bigger. You need to use the reciprocal of the IS peak to correct the analyte peak.

If you add 10% of diluent to 90% of sample the IS peak will still be smaller with a higher sample:diluent ratio.

A better way would be to do the dilution with clean solvent, then add a known mass of IS in as small a volume as practicle - then analyte and IS peaks both get smaller in the same ratio with an increase in diluent:sample ratio.

An even better way is to check weigh after addition of the diluent and correct for the actual amount added.

Peter

Peter,

Thank you for your reply.

If I understand correctly you make up a solution of IS in solvent, and then use that solution to dilute the samples.

No I am doing the opposite of that. IS is in the extraction solvent.

A typical dilution would go like this:

Take 100uL of sample extract (containing known concentration of IS 100ug/mL) and add that to either 400uL clean solvent for a 5x dilution or 900ul solvent for a 10x dilution.

An even better way is to check weigh after addition of the diluent and correct for the actual amount added.

I'm not sure what you mean here?

I'd say you could give positive displacement pipettes a try (Gilson and others manufacture them). They are perfect for handling liquids with a lot of vapour pressure such as alcohol.

sorry, I'm a bit sleepy and not with it today, but if you already have an internal standard in your sample extraction, is there any particular reason why you don't just set up your quantification method to use internal standard calibration instead of external standard calibration, make up the calibration standards appropriately (i.e. include the internal standard in all your calibration standards, at constant level), and let the software do everything for you?

Peter,

Thank you for your reply.

If I understand correctly you make up a solution of IS in solvent, and then use that solution to dilute the samples.

No I am doing the opposite of that. IS is in the extraction solvent.

A typical dilution would go like this:

Take 100uL of sample extract (containing known concentration of IS 100ug/mL) and add that to either 400uL clean solvent for a 5x dilution or 900ul solvent for a 10x dilution.

Then an increase in diluent volume will make the IS peaks smaller in the same ratio as it makes the analyte peaks smaller - just divide analyte area by IS area - which is what lmh has enquired about

An even better way is to check weigh after addition of the diluent and correct for the actual amount added.

I'm not sure what you mean here?

A quick and easy way to measure how much diluent you added is to weigh the vial plus extract, add diluent, and weigh again. You can then correct the analyte areas for any deviations from target addition

sorry, I'm a bit sleepy and not with it today, but if you already have an internal standard in your sample extraction, is there any particular reason why you don't just set up your quantification method to use internal standard calibration instead of external standard calibration, make up the calibration standards appropriately (i.e. include the internal standard in all your calibration standards, at constant level), and let the software do everything for you?

This is how IS should work.

This will allow IS to correct for any errors in preparation after the addition of IS and for any deviations in the readings from the equipment caused by minute deviations in injection volume, detector fluctuations, room temperature fluctuations, ect.

Response Factor = (As x Cis)/Ais x Cs)

Where As=Area of Analyte, Cis = Concentration of internal standard, Ais = Area of internal standard and Cs = Concentration of analyte.

Keep Cis constant in all calibration standards and use the formula to find the RF of each standard, then you can either take the average or plot a calibration line from the Response Factors. To find concentration of analyte in sample, you use the same concentration of internal standard as the calibration standards, solve for Cs and plug in the areas and IS concentration and you get the corrected concentration for sample. This is usually done automatically by the software when you choose to calibrate using internal standard method.

Hey everyone thanks for your replies.

sorry, I'm a bit sleepy and not with it today, but if you already have an internal standard in your sample extraction, is there any particular reason why you don't just set up your quantification method to use internal standard calibration instead of external standard calibration, make up the calibration standards appropriately (i.e. include the internal standard in all your calibration standards, at constant level), and let the software do everything for you?

I would prefer to do this. But the problem is the standards for the compounds I'm analyzing are rare and rather expensive. They come pre-diluted in solvent 1mg/mL. So I can't make up a standard stocks with equal amounts of internal standard easily (by weighing together).

Perhaps I could add internal standard while doing external standard dilutions.

For example:
Going from 1mg/mL analyate stock to 0.1mg/mL with 1mL volume. Take 100ul stock, 890ul solvent and 10ul of internal standard solution at 10mg/ml which would give 0.1mg/ml analyate and 0.1mg/ml internal standard.

Would this work?

Hey everyone thanks for your replies.

sorry, I'm a bit sleepy and not with it today, but if you already have an internal standard in your sample extraction, is there any particular reason why you don't just set up your quantification method to use internal standard calibration instead of external standard calibration, make up the calibration standards appropriately (i.e. include the internal standard in all your calibration standards, at constant level), and let the software do everything for you?

I would prefer to do this. But the problem is the standards for the compounds I'm analyzing are rare and rather expensive. They come pre-diluted in solvent 1mg/mL. So I can't make up a standard stocks with equal amounts of internal standard easily (by weighing together).

Perhaps I could add internal standard while doing external standard dilutions.

For example:
Going from 1mg/mL analyate stock to 0.1mg/mL with 1mL volume. Take 100ul stock, 890ul solvent and 10ul of internal standard solution at 10mg/ml which would give 0.1mg/ml analyate and 0.1mg/ml internal standard.

Would this work?

This is how I would normally do it. If you make different concentrations for a calibration curve you will want to keep the internal standard concentration constant though. If you have a 0.01mg/ml, 0.05mg/ml, 0.1mg/ml for the calibration curve you would want to have 0.1mg/ml in all three of those, plus spike the correct amount into the sample prior to dilution so that once the final dilution is done it should also be at 0.1mg/ml internal standard concentration. This way you track the response of the internal standard across all of the standards, samples and any blanks you inject to monitor how stable the instrument is as well as correct for any error in dilution of the samples.