mystery air in H2 analysis

[rekuci]

I need to analyze for oxygen, nitrogen, and hydrogen in some mixtures. The setup is an HP 5890 GC with a Carboxen 1010 (0.32 mm x 30 m)capillary column installed in a packed column inlet, TCD, and injection by syringe through a septum. Argon carrier flow is 1 mL/min with 8 psi head pressure, makeup and reference are 1 mL/min and 14 mL/min respectively. Oven 25C, inlet 200C, detector 200C. This appears to work very well, the baseline is great, and I get chromatograms from a 4-5% standard containing H2, O2, N2, CO, CH4, and CO2 that might as well be straight out of a column catalogue.

My problem is in attempting an absolute calibration with pure hydrogen, which also has implications for the relative calibration of H2/air I want to do. There is always a large amount of air (shows up as a non-resolved double peak at 5.5 min, where hydrogen is at 3.5 min) when injecting what should be pure H2. I'm sampling the H2 in a fill setup from a cylinder of UPC hydrogen at a pressure of 30 psi. The setup is leak-free (I tested it, and it holds 30 psi for at least hours), and it is filled from the tank and then purged out with a vacuum pump 3x consecutively before the final fill. I'm then sampling directly from a non-leaking septum adapter fitting attached to a quick disconnect fitting, using a Valco sampling syringe (the type with the push-button valve at the bottom). I close the syringe valve before removing it from the septum, and don't open it again until injection.

I tried injecting different volumes of 'pure' H2, eg 10 uL and 5 uL - the peak for H2 halves but the air peak does not halve (although it does vary a little, like from peak area of 490000 to 390000). Using an absolute calibration of air, it appears that air accounts for at least 20% of the injected sample. What on earth is happening here, where could this be coming from? Is air coming through the inlet septum when I inject the needle? Surely the dead volume in the needle is way too little to account for this much air. Also, the peak area for hydrogen for consecutive injections of equal volumes seems to vary more than it should, for example 10 uL injections can give peak areas between 2.0E7 and 2.5E7.

Any help is greatly appreciated.

[rekuci]

A clarification, since there was a similar post right before mine...

I need to analyze for nitrogen and oxygen in actual samples, so I can't just ignore it and factor out! And I have a good sampling setup that should be pure H2 (if this ain't, nothing is), so the copious air must be coming from somewhere...

[MikeD]

This does not surprise me at all. My experience with syringe injection of gases is similar. If I inject 100 ul of sample of say pure helium I typically get entrained air equivalent to 1-2 ul no matter what purging I do, perhaps a tolerable problem, until you inject 10 ul or less. I try to use only pressurised GSV sample introduction only. OK for some people it is not always possible, but I have not yet found a solution to the entrained air problem in small gas syringes. I would like to because it's difficult to purge a GSV from something like a flexible Tedlar bag (nevertheless that is what I have to do).

[rekuci]

Thanks for your reply. At least I'm not the only one with the problem, but it looks like 5 uL of air at atmospheric instead of just 1-2! I asked the person here who did a similar analysis before me, and he says he never tried to analyze pure H2 samples, so it's likely he never noticed. I'll try increasing the volume a little, but there's only so high I can go with a capillary column.

This is really going to introduce a lot of unwanted error since the amount of air does vary significantly from injection to injection. The samples that I'll be analyzing for someone will be primarily hydrogen with only small amounts of air, and he wants it all precisely quantified.

Do you have any idea where the air, at least in your samples, is coming from?

[chromatographer1]

The problem lies in your methodology. When you pierce the septum you lose some carrier gas and introduce air. Some may come from leaking around the seal in the plunger of the needle.

Go, buy a sampling valve and your problems will go away.

If you want to do a proper analysis you have to pay the price, sorry.

best wishes,

Rod

[MikeD]

Rekuci

While waiting for your sampling valve (you really must get one) why not try a packed column and larger injection. Your problem won't go away but it should reduce.

[chromatographer1]

buy a valve with 0.030" ID ports, usually called a packed column valve.

Even for capillary it is a good idea.

With the right valve you can directly inject 0.2 to 2µL with an internal sample loop.

With 0.030" ID tubing you get 4.56µL per centimeter of sample loop.

Good luck,

Rod

[rekuci]

I'm afraid to mess with this now-working system (after so many months of problem after problem) and take the plunge toward installing the packed column...

Just called Valco and the valve that was recommended to me is a 6-port HPLC (?) valve, at least it says it's for HPLC when I look at the product number. I can get this with a 10, 15, or 20 uL sample loop. But I'm having trouble seeing how this will work. The sample still has to get from the container it's taken in (a small cylinder with some Swagelok valves and a quick disconnect fitting at the end) into the sample loop without entraining air during this process. I'm guessing the sample loop will need to be vacuum purged to provide an air-free continuum to your sampling line, so won't this require an 8-port valve? It was so long ago since I used the sample loop on a gigantic Gow-Mac GC for gas analysis, but I think this was an 8-port sampling valve system with vacuum purging.

[chromatographer1]

Above valve has sample loops from 2µL to 10mL volume and 0.030" (0.75mm) ID ports.

Manually actuated for economy.

Short sample Transfer line should be 0.020" or 0.010" ID 1/16"OD line to valve. Purge line slowly and consistently. (If you put heavy oil into purge vessel below and put vent line into the oil, you can watch sample bubbling through the system.)

Vent line from valve (0.020" ID) should be placed into small purge purge vessel (2mL vial with septum) with a vent line to atmosphere to minimize back diffusion of air into system.

I assume you can squeeze bag for a few pounds of pressure to push sample through sampling system. Purge thoroughly and you should have no sampling problems.

best wishes,

Rod

[rekuci]

I'm not sure if I'll have enough sample to adequately purge to get rid of air, but maybe it's enough with such small diameter tubing. The sample cylinder is no more than 10 mL, and a typical fill pressure is 30 psi, maybe it could be higher, but it will certainly vary from sample to sample due to the nature of the experiment. The sample cylinder is a sort of metal high-pressure vessel and I don't think I'll have any ability to force the sample through as there is no bag.

Could 8-port with vacuum purge work? In position 1, the entire sample line up to the valve before the sample cylinder could be vacuum purged, then flip to position 2 (turn counterclockwise) and open valve to fill the loop, then flip to position 3 (turn counterclockwise again) to introduce the loop sample plug to the carrier gas line. To make sure the purge is complete, one could flip between filling the sample loop and purging a few times before injecting.

[chromatographer1]

I don't see why it wouldn't work.

Just try to use minimum ID tubing.

Good luck,

Rod