Stray Peak: Some Beginner's Questions

[Saladsamurai]

Hello All! I am quite new here and to GC in general, so I hope you can bear with me. I will try to keep this post as concise as possible. I have organized it into 3 distinct sections: 1) my background, 2) what I am trying to accomplish with GC and 3) what problems I am facing.

1) My Background: I am a graduate student in the field of mechanical engineering. My work is in experimental combustion. As a mechanical engineer, my chemistry is dismal at best, but I try my best to keep up with it when required (like now). I have read the first several chapters of Basic Gas Chromatography by Mcnair so I have a general (and qualitative) understanding of the underlying principles of GC.

I am using a Varian GC 3800 with Star Workstation 6.2 for software in order to test gas samples. The method that I am using is one that was developed by a Varian technician who gave on-sight classes before my arrival in this lab.

2) What I am trying to accomplish: This should be simple, I believe. I have tank that contains a known mixture composition of N2/CO2 (86/14). I would like to simply give this gas to the GC and calibrate it such that when I run an analysis, it can i) compute the correct composition when I give it another sample from the N2/CO2 mixture (i.e. it tells me that the mixture is 86/14) and ii) compute the percent composition of nitrogen if I give it a sample purified air (i.e. it tells me that the sample is 79% N2 and there are some other unknown constituents too).

I use a miniature regulator to monitor the pressure of the incoming sample gas at the inlet and a rotameter to monitor the flow at the outlet. I have included at snapshot of the setup below.

3) The problems I am facing: Remember that I am not even qualified to call myself a 'novice', so don't hesitate to assume that I am just plain dumb and state the obvious.

The way I have been instructed to do this (by someone who is only slightly more qualified than me) is to first calibrate the method by injecting a sample and taking note of the retention times. I then put these retention times into the 'peak table' of the method along with the gas names and their respective percentages. I then run the method as 'type calibration'. I then run the method again as a 'type analysis' and the result is the following image:


As you can see, there is a small peak that should not be there (I think). The N2 is on the left and the CO2 to the right. I have run this method about 7-8 times in an attempt to 'flush out' whatever that peak is. The area counts for this peaks have dropped with each run but they are taking longer and longer to reduce now (we are around 3500 counts now). Moreover, my results are a bit off. It is reporting ~85% N2 and ~15% CO2 instead of 86/14 and I think that the stray peak is what is causing this.

So my major questions are:

1) Does my procedure for calibration seem reasonable? And more importantly, when I "take not of the retention times" I have simply been using the figure generated and putting the it in 'cursor mode' and hovering over the peak and writing down the times. But what time should I be recording? The time at the highest point of the peak? Or the time that the peak starts to form? Also, is there a more precise way of doing this?

2) What are your thoughts on the stray peak? Are there any common reasons I should start trouble shooting?

Thanks so much for reading and please let me know if there is more information you need from me.


Image of gas inlet/outlet

[tom jupille]

It's been a loooong time since I did GC, but a few observations:

1. Assuming you are using a thermal conductivity detector, different compounds have different response factors so that 86:14 composition will not necessarily give 86% / 14% peak areas. You need to validate the method by running different standards at different ratios in order to determine the relative response factors.

2. GC columns and detectors have a limited range. Overloading the detector will tend to give "flat-top" peaks and (obviously) incorrect results. Overloading the column will tend to give tailing or otherwise distorted peaks. A quick look at your chromatogram suggests that you may be overloading the column.

3. Interfering peaks are common. They can come from things like residual oil in gas cylinders. They can be a tremendous pain to clean out. As long as they don't overlap with a peak of interest, it's probably best to handle them the "old-fashioned" way (ignore them!).

4. As long as you have a good separation, small variations in the way you read the retention time should not be a problem.

[Saladsamurai]

Hello Tom and thanks for replying

It's been a loooong time since I did GC, but a few observations:

1. Assuming you are using a thermal conductivity detector, different compounds have different response factors so that 86:14 composition will not necessarily give 86% / 14% peak areas. You need to validate the method by running different standards at different ratios in order to determine the relative response factors.

Yes. I am using a TCD (sorry, I should have mentioned that ). I am not exactly sure what is meant by "validate the method by running different standards at different ratios in order to determine the relative response factors".

What exactly is a standard? And if you don't mind, what is this 'response factor' you speak of? I understand to some degree how a TCD works: its essentially a wheatstone bridge circuit and the resistance changes as a function of temperature. Different compounds have different thermal conductivity and hence the resistance will be different (and the output voltage). Is this where the response factor come in?

2. GC columns and detectors have a limited range. Overloading the detector will tend to give "flat-top" peaks and (obviously) incorrect results. Overloading the column will tend to give tailing or otherwise distorted peaks. A quick look at your chromatogram suggests that you may be overloading the column.

By "overloading the column" do you mean I am 'feeding it' my sample too quickly (i.e. my inlet pressure is too high)? I have the inlet pressure set to 5 psig. I am not sure what a typical pressure would be for a N2/CO2 sample?

3. Interfering peaks are common. They can come from things like residual oil in gas cylinders. They can be a tremendous pain to clean out. As long as they don't overlap with a peak of interest, it's probably best to handle them the "old-fashioned" way (ignore them!).

Good to know! Thanks!

4. As long as you have a good separation, small variations in the way you read the retention time should not be a problem.

Ok. I was also thinking of just running the method and viewing the results and using the retention time that is gives me there.

Thanks again!

[chromatographer1]

I am curious about the nature of the column you are using.

The two 'sharp' peaks might both be N2 from the valve switching. And the broad 'doublet' might be CO2 and water coeluting. If the second sharp peak is separated in time for the same time the valve is held in an inject position before returning to the load position, then you have a 'reasonable' double sample injection, the first from the large volume sample loop and the second from the dead volume in the opposite side of the switching valve.

Too little information to be sure, but something perhaps you can check out.

best wishes,

Rod

[Saladsamurai]

I am curious about the nature of the column you are using.

The two 'sharp' peaks might both be N2 from the valve switching. And the broad 'doublet' might be CO2 and water coeluting. If the second sharp peak is separated in time for the same time the valve is held in an inject position before returning to the load position, then you have a 'reasonable' double sample injection, the first from the large volume sample loop and the second from the dead volume in the opposite side of the switching valve.

Too little information to be sure, but something perhaps you can check out.

best wishes,

Rod

Hello Rod! Thank you for your input. I don't really understand enough about how this machine works to fully understand your post, but I think I might be getting the basic idea. Is this part of method at all helpful?



I am just using what was given to me, so I have no idea why those times are meaningful (I actually doubt that they are meaningful outside of the scope of whatever that person was doing with gases at the time) or what series and bypass mean.


Also, here's another important question I have: How should I be controlling the pressure/flow of my sample gas? As you see from the image in the 1st post, I have a regulator at the inlet and a rota meter at the outlet. Is this typical? To me it makes more sense to have the rotameter AND the regulator at the inlet. What good is monitoring the outlet flow? Especially since it changes when I inject the sample?

Thanks again!

[chromatographer1]

I don't understand what you are doing unfortunately from the info you have provided.

Try to learn what this information means that you have provided.

It appears that you are doing two injections, one with a column in series and the second with the column removed from the flow path (bypass).

The little peak might be an flow upset from the valve operating.

The retention times are determined from the tops of the peaks.

The meter shows that you are getting sample through the sample flow path (you are not out of gas).

The flow should be zero when you are ready to inject (sample is under ambient pressure).

best wishes,

Rod

[tom jupille]

I am not exactly sure what is meant by "validate the method by running different standards at different ratios in order to determine the relative response factors".

To "validate" means to demonstrate that a method does what it purports to do. In simpler terms, provide evidence that you are measuring what you think you are measuring.

The "response factor" is the proportionality between a dependent and an independent variable. Assuming a linear response, it's the slope of the calibration plot between the variables. Different compounds have different response factors, and you have to take that into account. The relative response factor is just what it sounds like: the ratio of the response factors for two compounds.

Let's say that you have two compounds, and the response factor for compound A is 2x that for compound B. If you do the chromatography of a mixture and get equal peak areas, does that mean your mixture is 50:50? (Hint: "no"; it's 33:67).

Soooo, what you have to do is to determine the response factors for N2 and CO2 so that you can "correct" the area ratios. At the same time, you can check to make sure the response is really linear (which it may not be if you are overloading.

[chromatographer1]

Now now now,

let's be nice.

The gentleman has a degree in the field of mechanical engineering.

"my chemistry is dismal at best"

Why should he even know how to spell chromatography? He at least read a good beginners' book.

best wishes,

Rod