RRFs from instrument to instrument

[WK]

Hi All,
Thanks for any replies in advance.
I have a list of RRFs created on one instrument.
For similar column and flow rates - would the RRFs be transferable from
one instrument to another?
Regards
WK

[Peter Apps]

Hi All,
Thanks for any replies in advance.
I have a list of RRFs created on one instrument.
For similar column and flow rates - would the RRFs be transferable from
one instrument to another?
Regards
WK

How similar, and how much discrepancy can you tolerate ? Are they different instruments from the same manufacturer, from different manufacturers ?, AND most important of all, way more important than columns and flow rates - do they have the same detector ?

How difficult would it be to run a couple of representative test mixtures to see if there are any unexpected differences ?

Peter

[Don_Hilton]

I'll go a step further than Peter's answer. We are careful about using relative resonse factors on the same instrument over time - thus we run check standards and periodic recalibrations. Best, if possible, measure them all again. If you can't do all, then a good representitive mix that will change subject to changes in inlet condition, column condition, detector condition, etc.

[WK]

Hello Don and Peter,
Thank you for the replies.
Sorry I should have clarified that I generated them on an Autosystem FID with 5 type column.
I guess its easy to check a few compounds when setting up with another GCs FID and see.
Unfortunately I don't have another FID to hand to test this.
It is not a legislative/qc analysis - only semi-quant - much better than reporting area%.
Regards
WK

[Peter Apps]

Hi WK

If I understand correctly you will be reporting peak area ratios (or something similar) without running a calibration for individual components ?

Under these circumstances you need to worry much more about differences in inlet discrimination between different instruments, and even between different inlet liners in the same instrument, than you do about the FID relative response factors.

Peter

[WK]

Sorry Peter :
I have calculated average RRFs from 3 levels (against an internal standard).
I have then calculated %s in test samples using the RRFs and adding the internal standard used in calculating the RRFs.
This works reasonably well to +/- 5% of actual % (e.g. 0.95 - 1.05% for 1.00%).
Which is fine for the application.
I was just curious how well they would transfer to different FIDs.
Regards
WK

[Don_Hilton]

If that different FID is on an instrument with a different inlet, beware. I have seen relative reponses change just by replacing an inlet liner in the same instrument. It the particular case where it was a problem for me, I was using a liner with glass wool, and inlet activity can vary dramatically depending on how the glass wool was handled and how long ago. Other inlet liners can give variation across compunds, but typically not as dramatically.

[Peter Apps]

Sorry Peter :
I have calculated average RRFs from 3 levels (against an internal standard).
I have then calculated %s in test samples using the RRFs and adding the internal standard used in calculating the RRFs.
This works reasonably well to +/- 5% of actual % (e.g. 0.95 - 1.05% for 1.00%).
Which is fine for the application.
I was just curious how well they would transfer to different FIDs.
Regards
WK

So you have a three-point calibration that you want to transfer to another instrument. Bad plan I would say, and to run 3 points on the new instrument seems like way less stress than worrying if the two instruments behave the same.

Peter

[WK]

Hi Peter/Don,
Thanks again for the replies.
I am following "Intelligent use of relative response factors in GCFID" in Chromatography Today May/June 2012.
I must stress this is not for legislative/qc work - it is only to replace %area as a guide to %w/w.
Only as an in-house guide for creating a start for a formulation that will be adjusted and modified within legislative boundaries.
e.g. component 1 is 1.00% area but in reality 0.05% w/w.

I take on board your comments on injector liners - I use my manually wool packed liners -
I have trouble with the focus type liners and only pack empty liners the way Perkin Elmer recommend.
I would love to buy off the shelf fully packed liners with wool in place !

Best Regards
WK

[Peter Apps]

Hi Peter/Don,
Thanks again for the replies.
I am following "Intelligent use of relative response factors in GCFID" in Chromatography Today May/June 2012.
I must stress this is not for legislative/qc work - it is only to replace %area as a guide to %w/w.
Only as an in-house guide for creating a start for a formulation that will be adjusted and modified within legislative boundaries.
e.g. component 1 is 1.00% area but in reality 0.05% w/w.

I take on board your comments on injector liners - I use my manually wool packed liners -
I have trouble with the focus type liners and only pack empty liners the way Perkin Elmer recommend.
I would love to buy off the shelf fully packed liners with wool in place !

Best Regards
WK

Hi WK

I am slightly puzzled - the RRFs for the FID itself are well within your required tolerances except for molecules with a large proportion of hetero atoms (O, Cl N etc) so % area should work for you without using a calibration. If you really need the calibration to get the tolerances that you need, then you also need to worry about differences between instruments, and that means that you need to run the calibration on both instruments. Otherwise just use peak area % (or ratios).

Of course if the total weight in your w/w fraction includes material that does not produce an FID peak then you have to use a calibration of some sort to convert your peak area to weight in order to do the calculation of w/w.

Peter

[WK]

Hi Peter,
The solvent used mainly is ethanol (sometimes with water) and makes the differences seen between %w/w and area%.
Regards
WK

[Peter Apps]

Hi Peter,
The solvent used mainly is ethanol (sometimes with water) and makes the differences seen between %w/w and area%.
Regards
WK

Hi WK

For sure you must not include the solvent in the the % area reports - just inhibit integration until the solvent peak is passed and see if the % areas for the peaks of interest are any closer to what you expect for the composition.

Peter

[WK]

Hi Don/Peter,
Thanks for the replies.
Have a nice weekend both.
Regards
WK

[CharapitsaS]

Hi Peter,
The solvent used mainly is ethanol (sometimes with water) and makes the differences seen between %w/w and area%.
Regards
WK

I have a dream…. in your situation ethanol may be used as Internal Standard.

Theoretical background and analysis of the accumulated experimental data about this new method is presented in the paper http://pubs.acs.org/doi/abs/10.1021/jf3044956
Our experience shows that variations of RRF (relative to ethanol) of volatile compounds in spirit products (alcohol drinks) does not exceed 10% from different GC (FID) even from different vendors.

Best regards,
Siarhei Charapitsa

[Peter Apps]

I wish you good luck in using ethanol as the internal standard, except for a very narrow set of analytical conditions and sample compositions you are going to need it.

Peter