Ethanol Analysis in Gas Chromatography

[Analisis]

Hello, I am new in this. Perhaps you can help me with this. We have two Gas Chromatographs in two destilleries. One is Agilent 6850 with FID and the other 6890 with FID. Both have a split/splitless inlet and a column J&W Scientific Innowax 19095N-123 30m x 0.53 mm x 1.0 um. We are checking the ethanol for some compounds, the quantity of the compounds encountered will determine the quality or classification of the alcohol. We are checking the following compounds:
Acetaldehido
Acetona
Acetato de etilo
Metanol ISTD (ter amyl alcohol)
Sec-Butil Alcohol
Alcohol Propílico
Alcohol Isobutílico
3-Pentanol
Acohol n-Butílico
Acetato de n-Amilo
2,4Dimetil 3-Pentanol
Alcohol Isoamílico
Alcohol n-Amílico
Furfurol
Both equipments are calibrated with internal standard method. But one, the 6850 has only three calibration points and the other four calibration points. How can I be sure of the accuracy of the methods that I am ussing? I use a control sample with known quantities of each of the components calibrated and run this control sample and check for the accuracy of the method. I calculate an absolute error % between the expected amounts and the measured amounts. The % error I get is the following in each equipment:
Name| Error % 6890| Error% 6850
Acetaldehido| 49.74| -43.24
Acetona| -1.45| -16.66
Acetato de etilo| 0.89| -21.89
Metanol| -0.26| -13.73
ISTD| 0.00| 0.02
Sec-Butil Alcohol| -0.17| -7.31
Alcohol Propílico| 0.44| -9.00
Alcohol Isobutílico| -0.80| -9.29
3-Pentanol| 0.67| -17.21
Acohol n-Butílico| -2.33| -8.11
Acetato de n-Amilo| -2.17| -7.50
2,4Dimetil 3-Pentanol| -1.72| -10.80
Alcohol Isoamílico| -1.76| -7.12
Alcohol n-Amílico| -2.30| -6.42
Furfurol| -2.17| -9.13

So you can see that the 6850 with only three points of calibration is less exact than the 6890 using 4 points of calibration. I think that what is needed is to migrate from the three point calibration table to the four point calibration table, but the management in the laboratory says that it doesn't matter and that we don't need to upgrade to the four point table because of the time involved in making an extra point in the table. They said that is sufficient to have an % error less than 10%. I don't know what is the %error normally accepted in Gas Chromatography. Can somebody please help me with this?

Thank you for your help. Regards, Adolfo

[WK]

Hi Analisis,
Are you using the same calibration standards at both sites or are they independently prepared at the different sites?
I note there is a consistent negative offset.

I would suggest that adding the extra level to make 4 levels won't make much difference to the problem- ideally 7 levels for statistics anyway.

WK

[Analisis]

Hi WK,

The standards are different in both sites each site prepares his own standard. They are slightly different, but I mean that the control sample should read same in both sites. We use internal standard, and the quantity of internal standard varies slightly also from site to site. In one site is 16.119496 and in the other is 16.08716.

What do you mean?

Adolfo

[Peter Apps]

Hi Adolfo

How well do the calibrations fit a straight line ?

Do the calibrations span the concentrations in the samples - in other words is the lowest point lower than the lowest sample concentration, and the highest point higher than the highest sample concentration ?

As I am sure that you recognise, you have a strong negative bias on the 6850 with the 3-point calibration.

As a trouble-shooting experiment, try swapping the standards between the two instruments and see how it affects the results.

Peter

[Analisis]

Hi Peter,

Thank you for your advice. We will try swapping standards.

The purpouse of this analysis is to monitor the compounds in the Ethanol. Sometimes the quantities are minimal, so that they go under the minimal calibration point. In that case we respond saying that the compound is bellow the concentration of the minimal calibration point. When it goes above the maximal point of calibration the quantity reorted doesn't matters it is obviously out of range.

So we would want that all our samples goes bellow the minimal point of calibration, because it means a better product.

The trouble now is that one of the plants has discovered that the amounts in the analysis are sometimes much higher than they expected, because the results are not so low as the GC tells them.

Regards,

Adolfo

[gpronger]

My experience with the FID, is that is is an impressively linear detector, so I am a bit surprised with you finding up to 40% error. Since this is the same compound on both instruments (Acetaldehido) I would be suspicious of the actual standard that is being utilized. You do also have generally more "error" on the one instrument, so the idea that there is an issue with the standards utilized on that system is fairly likely.

One of the approaches to be certain that your calibration is working appropriately, is to utilize calibration verification standard. This would be a standard, run after your calibration, prepared from different sources from those you used to prepare the actual calibration. This would give you information regarding whether what you purchased is what the vendor said it was, and whether the calibration solutions were prepared properly.

Also, with two labs, you could run a periodic "Round Robin" study. (OK, I have no idea why, in the US we call these things "Round Robins", but as best I can tell they have nothing to do with a reasonably attractive bird found around the US).

What a "Round Robin" is, regarding quality systems (and not the bird), is the analysis of an externally prepared sample with known concentrations of the analytes, analyzed by a group of participating labs. Though your would have only two labs participating, it would give information how well each lab was performing the analysis.

Greg

[gpronger]

My experience with the FID, is that is is an impressively linear detector, so I am a bit surprised with you finding up to 40% error. Since this is the same compound on both instruments (Acetaldehido) I would be suspicious of the actual standard that is being utilized. You do also have generally more "error" on the one instrument, so the idea that there is an issue with the standards utilized on that system is fairly likely.

One of the approaches to be certain that your calibration is working appropriately, is to utilize calibration verification standard. This would be a standard, run after your calibration, prepared from different sources from those you used to prepare the actual calibration. This would give you information regarding whether what you purchased is what the vendor said it was, and whether the calibration solutions were prepared properly.

Also, with two labs, you could run a periodic "Round Robin" study. (OK, I have no idea why, in the US we call these things "Round Robins", but as best I can tell they have nothing to do with a reasonably attractive bird found around the US).

What a "Round Robin" is, regarding quality systems (and not the bird), is the analysis of an externally prepared sample with known concentrations of the analytes, analyzed by a group of participating labs. Though your would have only two labs participating, it would give information how well each lab was performing the analysis.

Greg

[Analisis]

Hi,

It is good to know that there is such a place like this forum. I want to thank you and all the people that is helping with this issue.

Well I have a lot of things to do in order to fix this analysis. I want to say also that I have been using a split of 2.5:1, an injection of 2 microliters and a injection port temperature of 250 Celsius. So other college say to me that I should have a large vapor formation in the liner. I will try to lower the temperature in the injection port to at least to 200, and perhaps inject only 1 microliter in order to find what happens.

Other Analyst say to me that the standards that he is using degrades when they are in storage, so he thinks that the calibration of the equipment should be done only once every six months!! It seems to me a large amount of time. What do you think? He says that the standard that he is using could not be used frequently because it degrades. He says that for example one standard of methanol, for example that has a purity stamped of 99% after two weeks it has no more 99 but instead perhaps 98%, and so on. He says that he checks the purity of the standards injecting in the chromatograph and checking with a percentage report, the purity of the compound. I don't know if this a good method. What do you think? How can I check the purity of the compounds I receive?

Thank you very much for your help.

Regards,

Adolfo

[gpronger]

Adolfo,

Regarding instrument calibration, I tend to rely on the standards to be more stable than the instrument. For example, the methanol standard at 99% and then 98% after a period of time, if he is injecting the neat material, so long as he is protecting it from picking up moisture from the air, I would have tend to believe that his detector dropped a percent or two in response than methanol degraded.

We will typically store our standards in a freezer to minimize volitilization and degradation. When we prepare a new standard we will typically transfer them from the volumetric to regular 2 mL autosampler vials and use them one at a time keeping them (particularly the unopened vials) in a freezer.

This has helped a lot in maintaining our standards over time.

Greg

[Peter Apps]

Hi Adolfo

Not to put too fine a point on it , calibrating once very six months is nonsense. There are so many things on a GC that change with time (inlet contamination, deterioration of the column, detector drift etc) that you really need to be calibrating with every batch of samples. If your colleague is only cleaning his inlet when he does his six-monthly calibration it could explain the difference in results !

Although I doubt that it is the sole cause of the problem, very low split ratios are difficult for the instruments to maintain, and difficult to reproduce on different instruments. Provided that the samples and standards are made up in exactly the same solvent (in this case ethanol) I cannot see how it would produce a systematic bias but I would be tempted to go for a geniune splitless injection with 1 ul injected rather than such a low split ratio.

Peter

[Analisis]

Hi Peter,

What you say make sense to me. The standards are there to check the equipment and not the other way. However, how can you be sure that your standards are OK? What check can you run in laboratory?

I am working right now in this analysis. I have changed solvent. Instead of water/ethanol I am using just ethanol.

Also have lowered the temperature in the injection port from 250 to 200 Celsius. Injection volume is now 1 microliter instead of two microliters. The chromatograms have bettered.

I have tried only with four compounds, but will try with all the compounds.

Another thing is tha we want to switch to Hydrogen as carrier gas. We have used Nitrogen,but want to try with Hydrogen, because it is a better gas for chromatography and also because we can use a generator instead of cylinders.

The other thing that I one to tell is that we have changed column, because a peak has started to appear, a very large one, and when we injected in the other GC the peak was not there. So, we changed to a new column, and the peak dissappeared.

Another question is about the plates. I have checked the plates in order to decide if the column is good or not, but I have seen that plates in some cases increase with the time, but resolution decreases.

Why this behavior?

Regards,

Adolfo

[gpronger]

Adolfo;

See how things are going for you after the switch to hydrogen. You should see better overall chromatography. Nitrogen theoretically offers the best HETP, but, the curve of nitrogen is fairly steep which means that you have a fairly narrow optimal flow rate. Further, the optimum for nitrogen is at a fairly low flow rate which simply meaning you have longer GC runs.

The issue with the "narrowness of the curve" is that unless you have the flow set to the optimum, and are using a GC with electronic pressure control (to keep the flow rate at the optimum) you are very likely well off the optimal flow rate as you are temperature programming (the viscosity of a gas increase with temperature, or in other words, at the same pressure, as the temperature increases, the flow is slower).

Hydrogen has a much "flatter" curve with the optimal flow rate well above (at faster flow rates) nitrogen. The benefit of it being "flatter" is that if you're not at the "perfect" flow rate, the chromatography doesn't degrade significantly. The other aspect of the HETP curve is that, from the optimal point (lowest) they fall off faster to slower flows than faster flows. For hydrogen in particular, as you increase flow from the optimum, there is little degradation. What that means in practical terms, is that you want to err to higher flows than slower flows. So, what I typically would do on a non-EPC GC, is set the temperature to the max I'd be running, and set my flow there. At the lower temperatures the flow will be higher, but my HETP in that direction doesn't degrade as fast. I also gain in a faster run time.

Once you have the instrument switched, post what's going on, and I'm certain Peter and I will be able to help you optimize your instrument.

Greg