Chromatography Forum

I need to evaluate test sample and standard stability for a validation protocol that I am working on which uses HS-GC.

I am in the developmental stage of the protocol (meaning I am still at liberty to change anything I wish) and was thinking about how to evaluate the stbility of samples.

My first thought was to use MHE to calculate the concentration of analyte in the HS vial. As days went on, stock standard would be added and similarly diluted in new HS vials and MHE would be performed again and concentration calculated. Concentration values would be used to generate a recovery of analyte and thus indicate stability.

However, this is going to be moderately time consuming. I was wondering if someone else had a different spin on how I may be able to evaluate stability. The only requirments are that stability must be monitored using the same equipment and test method as is being validated.

Thanks in advance,

Shaun

There are a lot of issues based on matrix and analyte specifics. Headspace sample stability can only be demostrated when the matrix and analytes are carefully and exactly defined.

Since I worked with different amounts of analytes and different matrices it was better to use concentrated standards which were then diluted and checked against fresh standards to determine std stability prior to std preparation. This is unneeded if the std is made just prior to analysis work.

Matrix stability was only determined for a 8 hour period with spiked samples prepared at the same time.

If you are able to go beyond this outline I commend you. For my employer it was more important to do the work in a timely fashion (within 24 hours) than to hope that every matrix and analyte could be checked for long term stability and consistency of measurement.

best wishes,

Rod

Presuming that you can make up standards that closely mimic your samples (which you need to do for calibration in any case) or that you have plenty of sample available; divide a batch of standard or sample among say a dozen vials, analyse one immediately, another after an hour, another after a further hour etc etc. If the samples are stable the peaks will always be the same size (presuming that the analysis is repeatable and not drifting).

Peter

Presuming that you can make up standards that closely mimic your samples (which you need to do for calibration in any case) or that you have plenty of sample available; divide a batch of standard or sample among say a dozen vials, analyse one immediately, another after an hour, another after a further hour etc etc. If the samples are stable the peaks will always be the same size (presuming that the analysis is repeatable and not drifting).

Peter

Peter,

Luckily I am able to exacly metch the matrix of the samples and standards.

Normally I would simply assay the same standard preparation once over time. However, from what I have both personally experianced and read on the issue is that the first injection in any headspace analysis is the most variable. On well behaving standards I have noticed varibility upwards of 25% when performing single injections out of a vial, which is not acceptable for the application.

Though in a real test situiation I suppose that I am going to be dealing with the same amount of varibility in any given sample analysis. Which means it would probably be best to change the method from a static method to MHE.

I guess what I am asking is: I need to develop a HS method that gives concrete numbers for the results (10% varibility max most likely). Is there any other way to do this than using MHE?

Shaun

I have not experienced the variability of HS analysis that you have seen.

I have found that headspace analysis is more reproducible and accurate than I am able to make standards reproducibly. In other words, my introduction of error in making the standards is higher than the variablility of the method itself.

I am addressing using the method within a range of 1 ppm to 1% and liquid samples of 25 to 200 µL and solvents from methanol to pyridine-tolunene range of volatility.

Variability of samples is caused by several factors:

inconsistent sealing of vials due to crimpers misadjustment, uneven surface of vial openings, and inconsistent vial caps-septa,

retention of analytes from septa material,

chemical reactions from solvent interaction, air, and matrices,

active sites in the analyte pathway (especially seen in the FIRST sample analyzed).

I hope this information is helpful. I spent several years determining the factors of inaccuracy in pharmaceutical headspace analysis. I am still amazed at its accuracy when all the factors of error are addressed and minimized.

best wishes,

Rod

Shaun

I have not experienced the variability of HS analysis that you have seen.

I have found that headspace analysis is more reproducible and accurate than I am able to make standards reproducibly. In other words, my introduction of error in making the standards is higher than the variablility of the method itself.

I am addressing using the method within a range of 1 ppm to 1% and liquid samples of 25 to 200 µL and solvents from methanol to pyridine-tolunene range of volatility.

Variability of samples is caused by several factors:

inconsistent sealing of vials due to crimpers misadjustment, uneven surface of vial openings, and inconsistent vial caps-septa,

retention of analytes from septa material,

chemical reactions from solvent interaction, air, and matrices,

active sites in the analyte pathway (especially seen in the FIRST sample analyzed).

I hope this information is helpful. I spent several years determining the factors of inaccuracy in pharmaceutical headspace analysis. I am still amazed at its accuracy when all the factors of error are addressed and minimized.

best wishes,

Rod

Shaun

I have not experienced the variability of HS analysis that you have seen.

I have found that headspace analysis is more reproducible and accurate than I am able to make standards reproducibly. In other words, my introduction of error in making the standards is higher than the variablility of the method itself.

I am addressing using the method within a range of 1 ppm to 1% and liquid samples of 25 to 200 µL and solvents from methanol to pyridine-tolunene range of volatility.

Variability of samples is caused by several factors:

inconsistent sealing of vials due to crimpers misadjustment, uneven surface of vial openings, and inconsistent vial caps-septa,

retention of analytes from septa material,

chemical reactions from solvent interaction, air, and matrices,

active sites in the analyte pathway (especially seen in the FIRST sample analyzed).

I hope this information is helpful. I spent several years determining the factors of inaccuracy in pharmaceutical headspace analysis. I am still amazed at its accuracy when all the factors of error are addressed and minimized.

best wishes,

Rod

Shaun

I have not experienced the variability of HS analysis that you have seen.

I have found that headspace analysis is more reproducible and accurate than I am able to make standards reproducibly. In other words, my introduction of error in making the standards is higher than the variablility of the method itself.

I am addressing using the method within a range of 1 ppm to 1% and liquid samples of 25 to 200 µL and solvents from methanol to pyridine-tolunene range of volatility.

Variability of samples is caused by several factors:

inconsistent sealing of vials due to crimpers misadjustment, uneven surface of vial openings, and inconsistent vial caps-septa,

retention of analytes from septa material,

chemical reactions from solvent interaction, air, and matrices,

active sites in the analyte pathway (especially seen in the FIRST sample analyzed).

I hope this information is helpful. I spent several years determining the factors of inaccuracy in pharmaceutical headspace analysis. I am still amazed at its accuracy when all the factors of error are addressed and minimized.

best wishes,

Rod

Shaun

I have not experienced the variability of HS analysis that you have seen.

I have found that headspace analysis is more reproducible and accurate than I am able to make standards reproducibly. In other words, my introduction of error in making the standards is higher than the variablility of the method itself.

I am addressing using the method within a range of 1 ppm to 1% and liquid samples of 25 to 200 µL and solvents from methanol to pyridine-tolunene range of volatility.

Variability of samples is caused by several factors:

inconsistent sealing of vials due to crimpers misadjustment, uneven surface of vial openings, and inconsistent vial caps-septa,

retention of analytes from septa material,

chemical reactions from solvent interaction, air, and matrices,

active sites in the analyte pathway (especially seen in the FIRST sample analyzed).

I hope this information is helpful. I spent several years determining the factors of inaccuracy in pharmaceutical headspace analysis. I am still amazed at its accuracy when all the factors of error are addressed and minimized.

best wishes,

Rod

Shaun

I have not experienced the variability of HS analysis that you have seen.

I have found that headspace analysis is more reproducible and accurate than I am able to make standards reproducibly. In other words, my introduction of error in making the standards is higher than the variablility of the method itself.

I am addressing using the method within a range of 1 ppm to 1% and liquid samples of 25 to 200 µL and solvents from methanol to pyridine toluene range of volatility.

Variability of samples is caused by several factors:

inconsistent sealing of vials due to crimpers misadjustment, uneven surface of vial openings, and inconsistent vial caps-septa,

retention of analytes from septa material,

chemical reactions from solvent interaction, air, and matrices,

active sites in the analyte pathway (especially seen in the FIRST sample analyzed).

I hope this information is helpful. I spent several years determining the factors of inaccuracy in pharmaceutical headspace analysis. I am still amazed at its accuracy when all the factors of error are addressed and minimized.

best wishes,

Rod

Hi Shaun

You can expect to get much better than 25 % repeatability from equilibrium HS, unless there is somethig very odd about your samples. Out of the box the Varian Genesys, Gerstel MPS2 and Agilnet G1888 samplers that I have used have given rsds of 1 - 3 %. Fiddle a bit with conditions and you can get it just below a percent. Go mad with optimisation and working around the design faults and rsds can be forced down to 0.2. - 0.3 %.

Rod - what set-up are you using, I am always looking for better hardware.

Peter

Peter,

I have used the Varian (Tekmar 7000) and the PE HS40 successfully.

When trying to measure 0.5ng of analyte in a vial I was forced to use smaller vials, 3.35mL were the prototypes. I was happy at <5% RSDs at that low level. PE later adopted 2mL vials in part due to my recommendation.

I concur with your numbers, although less than 0.5% RSDs are VERY COMMENDABLE. You really have optimized your analysis. If I got numbers less than 1% RSD I was satisfied, especially if higher concentrations were involved (more difficult to do reproducibly at 1% RSD I found) and time was an issue (minimal heating and analysis time was desired ( ! we got hundreds of analyses to do, git 'er done NOW ! )

best wishes,

Rod

Peter,

I have used the Varian (Tekmar 7000) and the PE HS40 successfully.

When trying to measure 0.5ng of analyte in a vial I was forced to use smaller vials, 3.35mL were the prototypes. I was happy at <5% RSDs at that low level. PE later adopted 2mL vials in part due to my recommendation.

I concur with your numbers, although less than 0.5% RSDs are VERY COMMENDABLE. You really have optimized your analysis. If I got numbers less than 1% RSD I was satisfied, especially if higher concentrations were involved (more difficult to do reproducibly at 1% RSD I found) and time was an issue (minimal heating and analysis time was desired ( ! we got hundreds of analyses to do, git 'er done NOW ! )

best wishes,

Rod

Peter and Rod,

You are getting that kind of accuracy out of single injections from headspace vials?

If that is indeed the case, then yes I have to agree that something has gone terribly wrong that needs to get sorted.

My setup should not be anything too odd as it is a stock PE HS40/Trap.