Unknown % recovery

5 posts / 0 new
Last post
Chricoli
Chricoli's picture
Unknown % recovery

When using eazy chrome elite on a Agilent 7890A and using an EC wax column, I am constructing a simple method using 5 levels of IPA and DI Water as external stds.  I am able to get a linear fit type of 0.999 and greater when constructing the calibration table.  When I re-run the level samples as an unknowns (same vials), my % recovery is up to 2% off of what it was when it was run as an external calibration std.   you can also tell in the reports that the peak area calculated for the samples, when run as unknowns, are at least 3-400,000 less than when those same samples are run as external stds. The integration is exactly the same as are all other parameters in the method. 

Would it be acceptable to run the samples as external stds to show system suit passes and then re-run the middle level external std and unknown samples together to do a one point calibration update of the original table?  This would slightly change the fit.

Also, does anyone have any better suggestions as to what kind of column to use when testing product with water and IPA in it.  I heard DB-1 are good.

Dr. Analytical
Dr. Analytical's picture
Can you give us some more

Can you give us some more information?
Column length, diameter, film thickness
Operating conditions: injector, oven, detector
What is your solvent?

This sounds like it could be an equilibration or system drift problem.  How many replicates of the standards did you run?  You might try making a series of injections of one standard and watching the area.

A DB-1 column would not work well for this analysis - it is a non-polar column and you have polar analytes.

Let's see if we can find a problem with the GC before we make any big changes.

Chricoli
Chricoli's picture
Column: 30 mm length 0.32mm

Column: 30 mm length 0.32mm dia 0.25um film thickness EC Wax
autoinjected via tower
oven: 60C 1 min 25C/min to 180C 0.3 min hold. Am able to achieve <1%rsd for each level of 6 injections during table construction. Am using "average relocates" when creating calibration point for table. Level one peak area between 1.5-1.27.million and the samples are around 1.09-1.11 million.

Chricoli
Chricoli's picture
Ops is the solvent peak of

Ops is the solvent peak of interest 82.4 C bp.

Dr. Analytical
Dr. Analytical's picture
There may be a long term

There may be a long term problem with stability.  You would have to try doing a series of injections of the same standard over the course of a day to see if there is a gradual change.  Try to duplicate the time change between your initial calibration experiments and the sample analysis.