Ethanol carryover problem with headspace

[mazeroth]

We do an annual calibration on our GC's at work and most have a headspace connected to them. We have Agilent 6890's with mostly 7694 headspaces (a couple G1888). We use Agilents protocol for the headspace: 5m x 0.1mm column with no phase, 200ºC isothermal oven, small glass capillary tubes that you angle in ethanol to suck it up and placed in a 20mL headspace vial. Headspace is set for 90ºC oven, 110ºC loop and 115ºC transfer line. The headspace used in this discussion is an older Agilent 7694.

Well, a fellow chemist said that a former chemist used to just put a few drops of Ethanol in the vials and not bother with the capillary tubes. I decided to give it a shot and did two test injections. The peaks were enormous and fat and about 5x larger than what we typically get so I decided it wasn't a good idea. I prepped the 6 test vials followed by a blank (to measure carryover, which must be < 0.1% to pass) and ran them. I got 4.2% RSD (limit is 2%) and 14.8% carryover! I then ran another blank and the carryover was smaller but not much. I then had an idea...

I set the GC to run for 15 minutes and I went into manual control on the headspace. Every 10 seconds I would open the sample valve, which puts the carrier flow through the sample loop, the transfer line and into the GC inlet. Well, every 10 seconds I get a very large peak, almost as big as the one in my carryover injection. I repeated this about 20 times; open the valve for 10 seconds, close it for 10 seconds (to allow the ethanol to accumulate) and each time I get a peak that is still huge but decreasing in size. Even when I did it every 5 seconds I would get a large peak. It's like the ethanol is endless!

This is the first time I've ran into a carryover problem so I decided to run Agilent's steam cleaning procedure, which has you inject 20 water vials (1mL water into 20mL vial) to clean out the flow path. After cleaning I ran 6 new ethanol vials and a carryover. I got 3.2% RSD this time and a much better 4.6% carryover. I ran the steam cleaning again! This time I got 1.5% RSD and 3.2% carryover. Better, but still 32x the allowed carryover!

We run a lot of methods that use DMSO and DMA, if that matters. I can't believe that me running the highly concentrated ethanol test vials in the beginning would have caused this but who knows. Does this sound like a contaminated loop or transfer line? Is there any way to decontaminate/deactivate it? Just replace?

RECAP:

First run after high concentrated test injections: 4.2% RSD 14.8% Carryover
After first steam cleaning: 3.2% RSD 4.6% Carryover
After second steam cleaning: 1.5% RSD 3.2% Carryover

Thanks in advance for any help! It's greatly appreciated!

[Johnny Rod]

Have you double-checked the carier gas flow through the HS sampler? It sounds like the ethanol vapour isn't being cleared out. What flow do you usually use, and how is the GC inlet set up?

[mazeroth]

Carrier flow is great. The system is set up with a transfer line directly plumbed into the S/SL inlet. 25 psi constant pressure, 918:1 split ratio (per Agilent), which equals ~0.2mL/min flow.

The chrom looks like it always does and the EtOH retention time is where it should be. What is crazy is the fact that I can manually open and close the sample valve and I get the EtOH peak. The peaks I get are roughly 5-6% the area of the injected sample. The one time I open and closed it 17x and the peaks were almost the same size each time. That's enough area to equal a single injection! Here's what that looks like:

http://www.flickr.com/photos/10690156@N02/6076829440/

Thanks!

[Johnny Rod]

The GC side will work fine I'm sure, the S/SL inlet will make sure the right amount of gas is going down the column, even when the gas flow from the headspace is way off. Can you measure the volume flow from the headspace (at the end of the transfer line) rather than the inlet pressure? If there is a blockage in the HS somewhere, the pressure will be maintained but the flow will not.

[Peter Apps]

Carrier flow is great. The system is set up with a transfer line directly plumbed into the S/SL inlet. 25 psi constant pressure, 918:1 split ratio (per Agilent), which equals ~0.2mL/min flow.

No matter where it is in the system; through the column or out of the headspacer, 0.2 ml/min is VERY slow and creates all sorts of possibilities for back diffusion and incomplete purging of sample loops.

Peter

[Johnny Rod]

The 0.2ml/min is down the analytical column in the GC i.e. after the S/SL inlet, it's a 0.1mm column so flow is okay. Maz flows are the thing here not pressures, and I'm sure the flow through the GC is just fine. The carrier flow through the HS is in question to me. Typically you'd set it a little lower than your total GC flow (column+split) for highest sensitivity, though it doesn't sound like your method needs any great sensitivity. At least you should have it set high enough to flow several times through the HS injection loop while it is in the "inject" position. The GC inlet will make up any shortfall in gas required from ths HS (your total flow is around 200ml/min so if the HS provides 100ml/min the S/SL inlet will add another 100), equally if the HS flow were too high the GC would try to dump the excess gas out of the split vent.

So anyway, rambling a bit, check the gas flow through the HS would be my suggestion.

[Peter Apps]

[quote="mazeroth"]Carrier flow is great. The system is set up with a transfer line directly plumbed into the S/SL inlet.
quote]

OK, so you have 200 ml/min through the loop and transfer line and 0.2 ml/min down the column. Can you explain what "directly plumbed" means - is this the set up where the carrier gas feed line to the inlet is cut and the transfer line connected into it ? If it is you very likely have a cold spot in undeactivated metal tube.

Peter