RRFs from instrument to instrument

[dblux_]

Unless there is something odd about the vertical scale that ethanol peak is the classic shape for column overload and detector saturation - therefore its area should not be used as a standard for other components.

As can be seen Y-axis is in log scale. Nevertheless in linear scale peak is strongly fronting due to massive column overload.
Increasing split ratio to reduce overload by EtOH in turn would affect congeners LOD.
That's why EtOH shouldn't be choosen as IS for official spirits drinks control with satisfying accuracy

[Peter Apps]

So there is something odd.

Odd also that "Ig" means log

Peter

[Peter Apps]

So there is something odd.

Odd also that "Ig" means log

Peter

[antonk]

So there is something odd.

Odd also that "Ig" means log

Peter

http://en.wikipedia.org/wiki/Logarithm#Particular_bases

[antonk]

Increasing split ratio to reduce overload by EtOH in turn would affect congeners LOD.
That's why EtOH shouldn't be choosen as IS for official spirits drinks control with satisfying accuracy

Amount ~ Peak area, not peak shape but Area which is proportional to ions count.

Unless your GC system drops substances through holes - what injected will go through column to detector.

[Peter Apps]

Increasing split ratio to reduce overload by EtOH in turn would affect congeners LOD.
That's why EtOH shouldn't be choosen as IS for official spirits drinks control with satisfying accuracy

Amount ~ Peak area, not peak shape but Area which is proportional to ions count.

Unless your GC system drops substances through holes - what injected will go through column to detector.

But if the peak shape shows certain types of distortion it as a symptom of detector saturation and then area (no matter how you measure it) is not proportional to quantity. And this has nothing to do with "holes" in the GC system.

Peter

[antonk]

But if the peak shape shows certain types of distortion it as a symptom of detector saturation and then area (no matter how you measure it) is not proportional to quantity. And this has nothing to do with "holes" in the GC system.

Peter

This is true only for detector which can be overflown:
I.e. flow-dependent detectors like TCD or for charge-coupling detectors like CCD in photometer.

Btw EthOH peak distortion is caused by column phase but not detector.

FID is just ionization chamber with up to 10^8 linear dynamic range.
FID linearity limited by ion recombinations only (assuming constant ionization effectiveness by H2 flame),
so you can numerically estimate N ionized/N total. Not sure such plasma density can be achieved in FID, it is not ITER.
It can be proven theoretically (Saha-Boltzmann equation) and practically (just dilute 2-component mixture and test).

[Peter Apps]

Every detector has an upper limit above which its response is no longer proportional to quantity.

8 orders is an extremely optimistic estimate of the working range of an FID on a GC - because the lower end of the range is set by what you can get through a column as a recognisable peak, not the LOD of the detector. Quoting from my post on Fri 13th (!) "The lower LOD for an FID is around 0.01ng in a peak, its linear range is no more than six orders of magnitude (which is a hundred times better than any other GC detector) which means that the ethanol peak in our generic beverage is dangerously close to the top end of the detector's linear range".

"FID is just ionization chamber with up to 10^8 linear dynamic range." is an unrealistic simplification. There is a lot of gas phase chemistry going on in a flame, and no one theory of its operation can accurately model its responses.

Peter

[dblux_]

This is true only for detector which can be overflown:
I.e. flow-dependent detectors like TCD or for charge-coupling detectors like CCD in photometer.

Btw EthOH peak distortion is caused by column phase but not detector.

FID is just ionization chamber with up to 10^8 linear dynamic range.
FID linearity limited by ion recombinations only (assuming constant ionization effectiveness by H2 flame),
so you can numerically estimate N ionized/N total. Not sure such plasma density can be achieved in FID, it is not ITER.
It can be proven theoretically (Saha-Boltzmann equation) and practically (just dilute 2-component mixture and test).

Several posts earlier Siarhei attached "Table 2" with a raw of numbers. Data for EtOH may be used to present linearity of response vs concentration:
http://i.imgur.com/CHusdZK.png
If I understand this table correctly graphical presentation of FID response vs. EtOH concentration is not linear for concentration range ca. 10^4.

I hope Siarhei will comment here if the right numbers have been chosen for the graph.

[Peter Apps]

Another serious problem with the ethanol as internal standard approach is that the errors are not random, they are biased and so you cannot get a closer approximation to the "true" value for an analyte by running multiple replicates.

Peter

[CharapitsaS]

Several posts earlier Siarhei attached "Table 2" with a raw of numbers. Data for EtOH may be used to present linearity of response vs concentration:
http://i.imgur.com/CHusdZK.png
If I understand this table correctly graphical presentation of FID response vs. EtOH concentration is not linear for concentration range ca. 10^4.

I hope Siarhei will comment here if the right numbers have been chosen for the graph.

Several posts earlier Siarhei attached "Table 2" with a raw of numbers. Data for EtOH may be used to present linearity of response vs concentration:
http://i.imgur.com/CHusdZK.png
If I understand this table correctly graphical presentation of FID response vs. EtOH concentration is not linear for concentration range ca. 10^4.

I hope Siarhei will comment here if the right numbers have been chosen for the graph.

Thanks to dblux_ for the deep interest in this activity.

Briefly recall.
To demonstrate the reliability/robustness of the proposed method “Ethanol-as-IS” the standard ethanol-water (96:4) solution with initial volatile compounds concentration about 4000 mg/L (AA) was analyzed after dilution with water in the ratios 1:1, 1:9, 1:99, 1:1999 and 1:9999. Experimental results are presented in Table 2.

There were two "slips of the pen".

The first slip of the pen (error) I made in the Table 2 when I prepared material for paper Instead of the rounded value of the number "257841,7" to an integer value (for the sample C , dilution with water (1:9) ) I wrote in error number "2578412".

Dblux_ made (added) the second slip of the pen (error). Dblux_ reversed the value of response for samples B and C. This situation is presented in detail in the following figure.
.

As a result, "Ethanol as IS" method continues to strengthen its viability/robustness

Best Regards,
Siarhei

[dblux_]

...
The first slip of the pen (error) I made in the Table 2 when I prepared material for paper Instead of the rounded value of the number "257841,7" to an integer value (for the sample C , dilution with water (1:9) ) I wrote in error number "2578412".

Dblux_ made (added) the second slip of the pen (error). Dblux_ reversed the value of response for samples B and C. This situation is presented in detail in the following figure.
.

As a result, "Ethanol as IS" method continues to strengthen its viability/robustness

Best Regards,
Siarhei

OK - curve is linear in this range of concentration.

BTW - I entered resp. values from table 2 in order of appearance into Excel cells:
2825852
1277251
2578412
21427
2173
223
and then sorted them automaticaly with "sort ascending" command.
As a result I got following column:
223
2173
21427
1277251
2578412
2825852
Hence it wasn't my error. It was natural consequence of your typo transferred by Excel to the final worksheet

[CharapitsaS]

Dear dblux_,

all results of experimental researches from three different laboratories in numerical and in graphic forms are presented in recent publication http://arxiv.org/ftp/arxiv/papers/1406/1406.6358.pdf .

Let's step into the future.
There is the following effective, simple, easy and reliable way to test the proposed method “Ethanol as IS”. This way does not require additional material, financial, or time-consuming.
On page 5 of the Certificate of measurement “Whisky – Congeners CRM LGC5100” indicated the following organisations took part in the interlaboratory exercise:

Chivas Brothers Technical Centre (UK)
Glenmorangie Plc (UK)
La Tondema Distillers (Philippines)
J&B (Scotland) Ltd (UK)
Joseph E. Seagram & Sons Inc. (Lasalle Technical Centre) (Canada)
Joseph E. Seagram & Sons Inc. (USA)
Revenue Canada (Canada)
Robertson & Baxter Laboratory (UK)
Scotch Whisky Research Institute (UK)
State Laboratory (Ireland)
Tatlock & Thomson (UK)
United Distillers & Vintners International Technology Services (UK)
William Grant & Sons Distillers (UK)

In order to accelerate the full test method and subsequent acquaint the scientific community with the results listed laboratories may send files of measured chromatograms of “Whisky – Congeners CRM LGC5100”. These chromatograms will be processed and presented for scientific community for a subsequent decision on new method application in everyday practice.

Best Regards,
Siarhei

[Peter Apps]

Paradoxically, the ugly overloaded peak actually helps with the quantitation. By spreading the peak sideways it allows an increase in area without much increase in height, and reduces the impact of non-linearity near the top of the detector's range.

Peter

[CharapitsaS]

Dear Peter Apps,

let’s look around and see new areas of application of the proposed method.
There are the following international standards in “Food Control”:
ISO 10315:2000 – Cigarettes – Determination of nicotine in smoke condensates – Gas Chromatographic method
and
ISO 10362-1:1999 – Cigarettes – Determination of water in smoke condensates - Gas Chromatographic method
In accordance with ISO 10315 researchers should introduce heptadecan as IS and in accordance with ISO 10362 – ethanol as IS.
In accordance with proposed idea we may use the main component of solution 2-propanol as IS.
Experimental results are in figures below.


ISO 10362


ISO 10315

No bad.

Best regards,
Siarhei