Chromatography Forum

Hello All,

I just wanted to get some opinions on my peak shape for a low concentration peak (500 ppb) of CO. I notice there is some tailing at the end of the peak, but what I do not have a sense for is what an "acceptable" level of tailing is. I know one can't get every peak in a method to look totally perfect, but what is too much tailing to make something quantifiable? I know it's especially an issue when dealing with such low concentrations as these.

Here are some chromatograms of a standard. The first is an overall picture of what is going on in the FID:



(I know the peak at the 6.7 minute mark looks like more of a problem. That's CO2. I'm not interested in low levels of CO2 such as this, 500 ppb, so this is acceptable for me.)

And this is a zoomed in shot of the CO peak, followed by the more symmetric CH4 peak (also 500 ppb).



Is the CO reliably quantifiable? If so, how should I draw my baseline? For previous analyses, I've been using the integrator with Chemstation, but should I be setting my parameters on that a certain way?

And in case it's needed, here are my process conditions:

GC = Agilent 7890A
Mixture: 0.1% H2, 500 ppb CO, 500 ppb CH4, 500 ppb CO2, balance N2
FID flow rates: H2 = 45 ml/min, Air = 500 ml/min, makeup (argon) = 30 ml/min, detector temp = 200 C
Column flow rate (argon) = 5 ml/min
Oven program = -80 C for 3 min., 60 C/min to 105 C, hold 1 min.
Column = 60 m GS-GASPRO
Valve Box T = 150 C
Nickel Methanizer Catalyst = 375 C
Injection Method - Via 6-port gas sampling valve, 30 second injection time.
Flow path - from 6-port valve, through column, into TCD, into Ni-methanizer catalyst, into FID.

Thanks,
fergman

Given your setup I think you have a nice chromatogram.

I would integrate CO by letting the integrator recognize the beginning of the peak and then have the software fix a baseline at about 3.44 min. and set the end the peak at that time (first tick on the baseline past the 3.4 min mark.)

As your column and methanizer ages the tailing will increase. I would guess that following what I described above will minimize the error associated with the inevitable tailing.

I would run a std at 50 ppb and see if you still have a recognizable peak at that level.

Given the tendency for tailing with CO, especially using a methanizer where CO chemically reacts (with a small but visible delay in completely reacting with the catalyst) I think your system looks good. It is not perfect but I would not expect perfect. Even so it is pretty darn good.

If it meets your needs, BE HAPPY. Especially if you can get a peak at 50 ppb clearly enough to register an area count (to give you a LOD level). A regular and scheduled test of LOD would be recommended. Catalysts do lose their effectiveness over time.

best wishes,

Rod

Hi Rod,

Thank you for the reply! Definitely puts my mind at ease.

As I begin to recalibrate, I have another question. This is one that's bugged me for awhile and that I've never gotten a satisfactory response to.

As I alluded to in my original post, the balance of my standards is nitrogen. For the standard I originally referred to, that nitrogen concentration is basically 99.9%. Now what I've noticed in the past as well as now is that I'm getting more variation in the TCD's response for nitrogen than I'd like. Right now, in three injections I'm getting from 120.7% to 126.7% (this is based off an old calibration curve...in terms of number of counts, this is 720,589,108 to 755,831,242).

Is this amount of variation normal? If not is there something or things I can do to improve it?

My TCD operating parameters are:

-Temperature = 200 C
-Reference Flow (argon) = 25 ml/min
-Makeup flow (argon) = 10 ml/min

-My gas sample loop is 250 ul.

-All other specs in prior post apply.

Here is an example of a chromatogram of the standard from the TCD:



The little peak just before the giant N2 peak is the 0.1% H2.

Thanks!,
Fergman

It is customary in commercial industrial applications (due largely to the variance you are seeing) to calculate the balance peak by difference, not directly. There are many linearity issues, mostly dealing with the nature of this detector in its physical shape and design, and the variance of cooling the filament when it contains a large minority or even majority of a non-carrier gas composition.

For example, for example if you had a gas mixture of high purity, with two impurities of 1000 ppm and 100 ppm typically, and if you were seeing 3% variance on a 1000 ppm peak and a 5% variance on a 100 ppm peak, if you tried to directly measure the balance using a TCD you would typically see a 1% variance (99.89 to 100.89 %).

But if you measured the two gases: ranging from 970-1030 ppm and 95-105 ppm and calculated the balance by difference, you would get 99.8865 to 99.8925 %.

Now you tell me, would you prefer to report the balance as

99.887 to 99.893 % OR 99.89 to 100.89 % ?

Which is more truthful?

Now you asked: "is there something...I can do to improve it?"

Yes, decrease your sample size by 10 or 50x. But, wait, I already know your answer.... NO WAY.
and your answer is correct.

I hope I have been helpful.

best wishes,

Rod

Hi Rod,

Thanks again for the quick response!

Yeah, difference would be ideal, but the problem I have is a complex sample; at this point, I don't know everything in my gas stream. I believe I know in the high 90s%, but because I can't unequivocally say I know everything, I wouldn't feel comfortable publishing a paper with that as an underlying assumption.

I asked about the TCD because, like in your example, when I don't use the difference method, sometimes I go over 100% (sometimes to even 103%), which is a nightmare for quantitation.

So if I understand you correctly, is my variation in nitrogen mostly due to oversaturating the TCD? Beyond reducing sample size, there's nothing I can do to tighten my variation?

I remember when Agilent came and first installed the instrument, they switched out the stock gas sampling loop for the smallest one they sold, the 250 ul that we're currently using. I see that Restek goes down lower, all the way to 5 ul (specifically, 5, 10, 25, 50, and 100 ul). And yes, my gut response to reduction in sample size was "NO WAY." Haha . Mainly because I'm trying to detect very low ppm levels of CO. However, do you think there is a gas sample loop size smaller than 250 ul that would allow for both the desired CO detection and a non-saturated, and therefore more consistent, nitrogen peak?

Right now, my 1 ppm standard for CO in the FID is at about 150,000 counts. And my lowest H2 standard in the TCD of 0.1% is a bit over 7 million counts.

Thanks,
fergman

You wrote:

"However, do you think there is a gas sample loop size smaller than 250 ul that would allow for both the desired CO detection and a non-saturated, and therefore more consistent, nitrogen peak?"

No. I don't. This issue is more complicated than that. But if you want to make loops and assumptions (which I am sorry to say, are unfounded) then here are the values for 1/16 inch OD ss tubing internal volumes.

id----- µL/cm----- µL/in
0.005----- 0.13------- 0.32
0.010----- 0.51------- 1.29
0.020----- 2.03------- 5.15
0.030----- 4.56------- 11.58
0.040----- 8.11------- 20.59

It is easy to make your own sample loops.

1/16 inch tubing with 0.010 inch ID will hold 1.29 microliters per inch in length.
1/16 inch tubing with 0.020 inch ID will hold 5.15 microliters per inch in length.
1/16 inch tubing with 0.030 inch ID will hold 11.58 microliters per inch in length.
1/16 inch tubing with 0.040 inch ID will hold 20.59 microliters per inch in length.

If you don't know the complete and true content of your stream then you cannot give a measurement for nitrogen anyway.

You can state that the smaller components concentration, based on a known calibration std.

You are already making assumptions about the gas as no gas mixture behaves ideally. That is why they are called the IDEAL GAS LAWS.

Some gases come close to theory, but when you don't know,
then YOU DON'T KNOW.

Just state in any publication your assumptions and how you are calculating and you will not be criticized.

I hope I stated it clearly. You are out of the linear range of TCD measurement whenever you try to measure plurality concentration AND you should not make assumptions about the accuracy of mole volume content.

It is chasing wild geese to try to make any claims about a 90% balance when you don't know the effect of the unknown gases on the actual mole amount of nitrogen in your sample volume.

You seem very determined to claim an accurate measurement of components of an unknown or partially known gas mixture. I wish you could but the world doesn't work that way. All you can do is tell the scientific community how you are measuring the sample, and what the numbers are.

You might use air as a calibration std for your nitrogen content. But keep in mind it is only a comparison and not an exact measurement of your unknown gas mixture.

Keep in mind that your barometric pressure variation in your lab also affects your results. Then you have low pressure your results are low. And when you have high pressure your results will be high.

And I wish you the best,

Rod