How to Determine LOD of Impurity in Gas Sample?

[Terry]

Hello all,

I am trying to determine the Limit of Detection (LOD) for oxygen in Nitrogen. I do not have much experience on handling gaseous samples, the LOD determination is kind of challenging for me.

I have performed LOD study for liquid samples according to ICH Q2 before, that is pretty simple: just dilute a standard of known concentration to a certain degree until you get a Signal-to-Noise (S/N) of about 3, then the concentration would be the LOD.

For gaseous samples, I have no idea on how to dilute a gas standard (let's say 1% oxygen in nitrogen). Anyone could direct me a way to around this problem?

Thanks in advance,

Terry

[chromatographer1]

The simplest way, the quickest, cheapest way would would be do change the volume injected for the determination of the LOD of oxygen.

Inject 1mL and then inject less, all the way down to 1µL. A 1% O2 in Nitrogen might have 10000 counts for O2 with a 1mL injection. With a 1µL your counts might not be measurable, or it might be 10. A 10µL syringe can give you easily changeable volumes from 0.5 to 10µL.

Also don't forget to inject a blank (pure N2) just to determine if you have any O2 being injected from the very act of injecting or from a leaking syringe.

Good luck,

Rodney George

[Terry]

Thanks Rodney for your reply.

Our Agilent 7890A GC is equiped with a Gas Sampling Valve. The size of the sampling valve is fixed. Change of sample size is not an option here.

[chromatographer1]

A. You can plumb the inlet lines to allow manual injections. This is inexpensive and convenient. I have done this for 5890 models and the same can be done on later models.

B, Otherwise, new cylinders containing known mixtures of gas must be prepared. You have to construct a gas mixing station that allows you to evacuate a cylinder to vacuum and then fill it to a known pressure with your std gas mix and then add pure nitrogen to a known pressure.

I have done this but it requires the purchase of expensive valves and vacuum pumps as well as digital gas pressures gauges and lots of time checking out your system for leaks.

A. costs $20

B. costs $1000-3000 depending upon labor costs

Good luck,

Rodney George

[GasMan]

I would not recommend using a syringe for gas injection, especially for the project that you are trying to do. There will always be air in the syringe needle, so your oxygen values will always be wrong.

You can change the loop sizes on the injection valve. Valco supply small loop sizes.

Another approach is to inject your samples at sub ambient pressures, other wise you will have to take route B that Chromatographer1 has mentioned.

Gasman

[Terry]

Thank you both for your time.

Gasman, could me please direct me on "inject your samples at sub ambient pressures" in detail?

Terry

[GasMan]

I have never actually tried this, but it is documented in several GPA methods. You will need a vacuum pump, a pressure gauge that can measure absolute pressure, and valving to control the pressure. You would fill up the sample loop at atmospheric pressure, 760mm Hg. You then reduce the pressure, to say 380mm Hg in the sample loop, which would mean that if you had 1% oxygen in your sample, this would now be equivalent of injecting 0.5% oxygen.

By using different pressures you can inject different amounts.

I personally prefer to use the method B from Chromatographer1.

Gasman

[Terry]

Rodney & Gasman,

Thanks again. Now I have gotten the answer to my question.

Good day ( night here in China)!

Terry

[chromatographer1]

Gasman is correct if you use poor technique in sampling your gas manually, simply removing the needle from the sampling space and carrying the syringe through the atmosphere to the injection site as air will diffuse into the needle tip as your sample diffuses out into the atmosphere.

But if you sample an excess of volume before you withdraw the needle and then continue to slowly expel the sample as you approach the injection site you can make an accurate transfer of gas. Not easy necessarily, but it can be done with care and practice.

Sample loops are more easily replaced in a valve plumbed ahead of your GC's valve which is not readily accessed and getting down to very low volumes is difficult. This can also be done with a temporary basis with the value retained as a backup for your GC valve.

But a simple piece of tubing with a T-union with a septum cap is so much easier to use and cheaper as well. But that is my choice unless government reviewers are a part of the research, if so, then buy different mixtures of gas commercially and document your serial numbers of the cylinders and their concentrations.

best wishes,

Rodney George

[larkl]

Here's a way to do it manually. I use it all the time. It was shown to me when I was a co-op student (a long time ago).

Print out chromatogram including the peak in question. You need to know the amount of the peak, you can use a std or a quantitated sample. Use a ruler (draw lines to the left edge)and estimate the height of the peak, probably in mV.

Zoom in (WAY in) on the background near this peak. Zoom in until it looks quite noisy (maybe the noise is half the screen). Print this out. Use a ruler and measure the noise, also in mV.

Now it's just simple math.

LOD = 3 x Peak Concentration/Peak height x Noise Height.

[Peter Apps]

Nearly all software packages will do this signal:noise at the press of a couple of buttons. It is hopelessly optimistic as an estimate of LOD because it assumes perfect linearuty and a 0,0 intercept and ignores run to run and sample to sample variability.

Peter