irreproducible peaks (+/-1000%)

[rekuci]

I'm trying to analyze different compositions of hydrogen/air mixtures to perform a calibration, but have been getting some really insane results. The samples are gas-tight syringe injected into a packed column inlet adapted to a Carboxen 1010 capillary column, with argon carrier gas, no inlet purge, and a carrier flow of 1 mL/min (reproducibily always 8 psi head pressure). I always inject 10 uL of gas at 30 psig.

Raw peak areas and the ratio between the hydrogen and air peak are completely different with almost every injection. I'd be happy with 15% right now. I've tried: replacing the septum daily...running a cleaning wire through the needle after each use...Merlin microseal septum...different injection techniques...loosened the septum nut...different needles...column conditioning...nothing helps at all! What could be the problem?

This analysis was done successfully with no more than 15% variation a few years ago with an identical setup. Also, a month ago, I did separate air and H2 calibrations with different volumes of the pure gases, and this gave very clean slightly parabolic area vs. volume curves (didn't repeat each injection 10x though).

Thanks for any assistance.

[Earl]

What is your air/H2 mixture held in? Is it mixed in a tedlar bag or what? How are you making the mixture?

[Earl]

Just saw your last post about the valve. Are you still using this?

[rekuci]

I have this setup of Swagelok fittings, valves, and stainless steel tubing that I'm using to fill with H2 and air. I can read the pressure on a pressure gauge. There are valves going to a hydrogen tank and a compressed air tank. There is also a valve before a hose adapter where I can apply vacuum from a pump. I take the sample through a Swagelok fitting with a piece of Thermogreen LB2 septum jammed in the end, just puncturing through it with a syringe. At one point I verified the entire thing was leak-free. It will hold 30 psi for at least several hours, except when the septum is cored by too many punctures (where it'll slowly drop off to 28-29 psi over a number of hours). It will usually still be at 25 psi the next morning if left filled overnight.

So all I have to do to get the mixtures is vacuum purge the whole thing a few times, open the valve to the air tank and fill to desired psi, then open valve to hydrogen tank and fill to desired psi. I'm always using 30 psi mixtures right now, and sampling 10 µL.

As to the HPLC valve, yes, I still had it in the carrier flow line before the inlet. It had been leak-checked and appeared ok. I just left it in the position where carrier flows directly through one port, out another port, and into the inlet. But out of paranoia, I just now disconnected it.

Here are some examples of troubles I've been having, running various samples repetitively:

42.6% H2 (25.7 psi air, 19 psi hydrogen)
H2 area air area H2/air ratio %H2 of total area
757541 4745289 0.159640646 13.76638929
1196186 4755861 0.251518284 20.09705233
3468361 3616280 0.959096364 48.95605861
4900207 3661556 1.338285417 57.23362116
733531 4754489 0.154281775 13.36604094
82698 307626 0.268826432 21.18701387
1034601 1374030 0.752968276 42.95390203
9478186 3101694 3.055809503 75.34400964
1353458 4784340 0.282893356 22.05119817
4442774 4006212 1.108971268 52.58351712
5394391 3846325 1.40247925 58.37633144
7579108 3252987 2.329891881 69.96899492

67.1% H2 (14.7 psi air, 30 psi hydrogen)
H2 area air area H2/air ratio %H2 of total area
633833 2298069 0.275811127 21.61849202
6473828 3631960 1.782461261 64.06059577
8543276 3150831 2.711435809 73.05624961
12540105 2310381 5.427721661 84.44238795
379089 44917 8.439766681 89.40651783
496471 390462 1.27149633 55.97615603
12057101 2760894 4.367100294 81.3679651
12908567 2497216 5.169183202 83.79039871
11291239 2112355 5.345332106 84.24038359
14014221 2277706 6.152778717 86.01941931
15202767 2531683 6.005004181 85.72449103
8012844 3661509 2.188399373 68.63630044
8434407 3291794 2.562252377 71.92787332
13232967 2840578 4.65854731 82.3276197

79.1% H2 (9.2 psi air, 34.7 psi hydrogen)
H2 area air area H2/air ratio %H2 of total area
10551738 2072728 5.090749003 83.58165803
14533346 1977771 7.348346194 88.02157964
15855505 1901999 8.336232038 89.28904085
14810342 1653200 8.958590612 89.95841843
16610647 1574187 10.55188932 91.34340737
15479500 1371573 11.28594687 91.8606192

90.2% H2 (4.3 psi air, 39.4 psi hydrogen)
H2 area air area H2/air ratio %H2 of total area
17815222 1850980 9.624751213 90.58801491
12900338 862361 14.95932446 93.73407062
20673580 1198157 17.25448334 94.52189371
23347011 1075554 21.70696311 95.59606454
22612720 1108895 20.39212008 95.32538151

100% H2 (44.7 psi hydrogen)
H2 area air area H2/air ratio %H2 of total area
25105482 580089 43.27867276 97.74157639
21817959 694435 31.41828825 96.9153214
22722826 491720 46.21090458 97.88184529
25927819 114563 226.319309 99.56009016
26296912 448532 58.62884254 98.32295923
27314466 233192 117.1329462 99.1534961
26683224 507478 52.58006061 98.13363406
25348682 506430 50.05367376 98.0412771
16159251 459432 35.17223659 97.2354488
24856097 431914 57.54871803 98.29202067
21695140 439194 49.39762383 98.01577947
28014112 252543 110.9280875 99.10656921

A couple trends are:
-100% hydrogen is fairly reproducible, it seems to get worse with the mixtures.
-Over a course of several identical runs, there is a trend where hydrogen peak are will *increase* and air peak area will *decrease*. You can see that trend buried in some stretches of this data, but you don't have the order of the samples.

[Earl]

This is a real noodle scratcher. It seems like the more air in the mix the less reproducible the areas at the GC. Have you tried 100% air and checked repeatability there?

When you say that the identical setup had less than 15% drift, does that include the gas mixing setup?

I may not be much help here, but I would begin by convincing myself that the GC has good repeatability. Then I would begin to question the mixing apparatus. If you have access to a summa canister or a small cylinder you may be able to capture the gas mix in an evacuated canister to convince yourself that the gas mixing apparatus is not the problem.

Sorry to not be much help. If you contact the company Entech, they may be willing to help.

[rekuci]

It was a while ago, but I did a calibration with different volumes of air and it had come out looking nearly linear (so did a calibration with different volumes of hydrogen).

I haven't tried verifying with air recently though.

Yeah, I've suspected the mixing apparatus but I can't imagine what could be going wrong in it. It probably has a pretty small volume, but gases should mix quite readily, and if nothing is leaking...

We use the mixing apparatus to fill sample cylinders (which I think is also not reproducible), also this apparatus will be used to take real unknown samples from a bomb.

I'm pretty sure the identical analysis from a few years ago used the same mixing setup, but the person doing the analysis was provided with the samples in the sample cylinders filled with the setup. He was sampling by syringe through the same sort of Swagelok septum adapter, only I think he used 20 uL instead of 10 uL, but 20 uL hydrogen peak is so huge and wide (argon reference is REALLY sensitive to hydrogen) that you don't get good baseline resolution with the air peak a couple minutes later.