Frustrating air leak in new GC, help?

[MAForensics]

We recently had a brand new system installed, and Agilent 7890 paired with the 5977 MS with the new high efficiency source. But we have an N2/air leak that will just not go away and we've explored almost EVERY possibility I can think of.

Here's the rundown:

The N2 level is about 10% at a column flow rate of 1 ml/min. We are using ultra pure He as carrier. The N2 percentage INCREASES to about 30-40% when you increase the column flow from 1 ml/min to 2 ml/min.

The mass spec is not leaking. When you plug off the transfer line, there is no air/water present.

Everything in the oven is tight, using the new hand-tightened inlet/interface nuts from Agilent. All parts are Agilent including the column.

Liner/septum is installed correctly.

The entire flow path has been replaced, by Agilent. So EPC, inlet, split vent line, etc. No change. The ENTIRE GC was then replaced by Agilent. No change.

The filter installed with the system went bad within a week (saturated with O2, indicator turned completely black). Not surprising given there's air in the system. We assumed this meant out gas was bad. However, the entire laboratory is fed on a 4-tank system. No other GC/MS systems show any nitrogen leak.

The GC/MS are set up on benches in pairs. So the gas line splits 3 times after the tanks to feed 3 separate benches. We assumed this particular bench had a leak (previously not in use). We hooked the new GC up to one of the lines feeding one of our other GC/MS. No change.

So then we hooked up one of our GC/MS already in use to the gas line on the bench for the new GC/MS. No leak shows on the older system. Gas line is fine. To be extra sure of this, we rolled a fresh, new tank out and hooked it up directly to the EPC of the new system. Still leaking. This occurs with and without a filter/scrubber on the gas line. With a scrubber, the O2 amount is decreased (this should imply the air is entering the line BEFORE the scrubber, but we find no leak in the gas line).

Agilent performed a leak test on the inlet. Passed to their specifications.

We've changed the column, twice. As it turns out, the new finger tightening nuts use the same ferrules for inlet and interface, so we had the wrong ferrule in the interface nut. Fixed this. No change.

I'm at the end of my rope. Agilent is at the end of theirs. Any help or insight would be greatly appreciated. I have essentially taken this thing apart and put it back together and I still cannot find anything. The fact that the air leak increases with flow would imply it's somewhere in the flow path, but we've replaced that. Twice. And nothing.

[dblux_]

What is the ratio of N2 to O2 ? How do you know you have an air leak ?

[tkubowicz]

Hello

Have you got N2/O2 ratio as in air (78:21 %) all the time? Check ratio for different flows and oven temperatures to confirm that it is air leak.
Perhaps 28 ion comes from something else.

Regards

Tomasz Kubowicz

[MAForensics]

Without a scrubber on the gas line we are getting roughly the same ratio as atmosphere.

I'm willing to entertain the possibility of it NOT being air, but regardless the MS won't pass an air/water test or a tune eval and therefore we can't run it.

[Gizmo]

They replaced the ENTIRE GC! - that is some dedication by Agilent. Based on what you wrote my focus would be on the transfer line. I'll assume this was not replaced when the entire GC was replaced and it seems like Maybe this could be a source of the problem. I know that it plugged up OK but perhaps it has an issue when a column is in the mix. I've never actually done this so I apologize if this is a silly suggestion but would it be possible to move the interface assembly from one of the other GCMS units and give it a try?

[tkubowicz]

Hello

Even if you can't pass air/water check you can manually set up tune window and check ions intensity (abundance) when you change flows, temperatures etc.
If you passed air/water check with MS plugged, you can assume that MS works fine so you can use manual tune as diagnostic tool.
You can use argon to check leaks - spray possible leak places and observe MS spectra.

Please see link below: it is helpful article

https://www.agilent.com/cs/library/tech ... 3899EN.pdf

Regards

Tomasz Kubowicz

[Peter Apps]

How are you testing for leaks ?

Peter

[dblux_]

If the GC and MS are new then let service engineers do their job. When they finish ask for tests. Don't forget to ask for sensitivity check. You have already paid for it, haven't you?

[MAForensics]

We are testing for leaks using Snoop (soapy water, basically) on all the copper fittings, and a spray can of duster (DFE) on the GC points. Agilent has also gone over everything with a leak detector. Nada.

[MAForensics]

Hello

Even if you can't pass air/water check you can manually set up tune window and check ions intensity (abundance) when you change flows, temperatures etc.
If you passed air/water check with MS plugged, you can assume that MS works fine so you can use manual tune as diagnostic tool.
You can use argon to check leaks - spray possible leak places and observe MS spectra.

Please see link below: it is helpful article

https://www.agilent.com/cs/library/tech ... 3899EN.pdf

Regards

Tomasz Kubowicz

Thanks for the reply. We are currently using manual tune to watch 28 ion intensity. When I say we can't use it, I mean that even if it's NOT due to air leak (i.e. the 28 ion is some random contamination that won't effect our runs), our protocols specifically state that we must have a passing tune evaluation in order to run samples. Therefore, regardless of whether it's air or not, we need to get rid of it.

[Peter Apps]

Nasty one.

The exhausted scrubber, and the lack of effect of two new GC flow paths points to air in the helium as it enters the scrubber. Lack of air on the other systems points to clean helium in the cylinder. So there has to be a diffusion leak upstream of the scrubber element, and downstream of the T to the other systems. You find no leaks in the pipes or connections, how about the scrubber itself ?, a cracked weld, chafed O-ring, scratched sealing surface depending on design and construction ?

Also if you are having to disconnect and vent the gas lines to make the troubleshooting changes you describe you are introducing air every time, and with multiple branches you need to bleed the end of every branch at a high flow rate to be sure that you do not have a dead volume that diffuses air back into your gas stream.

I don't let snoop anywhere near anything with clean gas in it.

Peter

[MAForensics]

Nasty one.

The exhausted scrubber, and the lack of effect of two new GC flow paths points to air in the helium as it enters the scrubber. Lack of air on the other systems points to clean helium in the cylinder. So there has to be a diffusion leak upstream of the scrubber element, and downstream of the T to the other systems. You find no leaks in the pipes or connections, how about the scrubber itself ?, a cracked weld, chafed O-ring, scratched sealing surface depending on design and construction ?

Also if you are having to disconnect and vent the gas lines to make the troubleshooting changes you describe you are introducing air every time, and with multiple branches you need to bleed the end of every branch at a high flow rate to be sure that you do not have a dead volume that diffuses air back into your gas stream.

I don't let snoop anywhere near anything with clean gas in it.

Peter

So, it's hard to explain how our instruments are set up, I don't think I did a good job.

We have 4 helium tanks in a separate room hooked up to a regulator. From this regulator it gets piped across the drop ceiling and behind a wall. At some point (no idea where), it splits into 3 (actually 6, we use 3 and another unit uses 3). These three lines then come out of the wall at separate points to supply a bench (one line per bench, three benches all parallel to eachother). Each bench line then splits AGAIN to 2 'spigots'. The first two benches each have 2 GCs on them. The third bench was not in use and now has the new system.

If I hook up an old system to the gas line on the third bench (just run a copper line across the space between the two benches), there's no leak present in the older system. So this should tell me that the helium coming from that 'spigot' is fine, yes? But then we hook the new system up to the same line, and suddenly we have a leak. With or without the scrubber.

Originally we figured the same as you, that it had to be upstream somewhere in the bench or the wall since the scrubber was getting fried, but that seems to not be the case! We've tried two separate filters like these: http://www.agilent.com/en-us/products/l ... ter-system

Both of which fried with oxygen saturation in a few days. We're now using one of the large traps (similar to this, or might actually be this, not at work right now: http://www.chem.agilent.com/store/en_US ... navCount=7), and we assume it's still ok since the oxygen is near 0 at the moment.


We've also hooked up the new instrument to helium by T-ing off one of the older system. When we do the T, we have no leak in the older system, same leak in the new system. We've also run directly from the tank to the new instrument through a regulator and still show the leak.

So we're fairly certain it's not our gas or gas lines. But we're assuming it has to be somewhere near the EPC if it's diffusing into the filters.

I was really hoping there was going to be some crazy glaring change to these new systems that we might not have thought of (I was really excited to learn we used the wrong ferrules with the column install, but alas this morning the 28 ion was still there and annoying as ever).


Edit* Each of these different hook ups has been tested with and without a filter and with helium running (i.e. fast changes without shutting the gas off and getting dead volume). Each change we let sit for a day or two to make sure it's not just residual air.

We have had ONE instance in which we managed to get rid of the 28 ion and pass a tune eval. We thought we had it fixed. The instrument sat, untouched, unchanged, for 4-5 days and when I went to use it again... 28 ion was back. We've been unable to duplicate whatever we did.

The real issue is we have no other 5977 with the HES to compare to, but even the guy from Agilent confirms there shouldn't be anything about the system that would make it more susceptible to leaking or that would throw some sort of false 28 ion.

[Peter Apps]

It looks like you probably have multiple problems, and the 28 ion will only go away if you fix all of them at the same time - a rare exception to the change one thing at a time rule.

As a double check that there really is air (as opposed to something else) in the helium as it enters the scrubber, check back through the tests to see if you had oxygen as well when you took the scrubber out, and when the scrubbers were saturated.

As a check that the 28 is not coming from the column (this is a recurrent possibility with mysterious "air" leaks and I had one case where the column was the only source of 28 that I could not eliminate) swap the column for another with a completely different stationary phase, so if you are running the default 5% phenyl siloxane put in a wax, or preferably an uncoated silica tube. Do this while your current scrubber is still taking the oxygen out.

There are a LOT of connections in that gas system, and no way of leak checking those in the wall. The absence of 28 on the older systems still says no leaks though, but for the present I would stick with a single cylinder feeding only the new machine. NB that there are several ml of air inside the Borden tubes of a regulator (certainly enough to saturate a scrubber) and it is very difficult to get it out because they are blind ended. Once you have the regulator on the cylinder, with an on-off valve between regulator and pipe and the end of the pipe open to atmosphere cycle the pressure in the regulator up and down a dozen times by slightly opening and closing the tap on the cylinder - this pumps helium into and out of the Borden tubes which cleans out the air much faster than diffusion would. Then, with the helium running gently close the valve downstream of the regulator, set the line pressure, then open the valve to purge the line, connect to the line to the scrubber with helium flowing. Now that the regulator is hooked up and clean keep it pressurized all the time.

Peter

[Bigbear]

Has this manifold in question been in service ever? It may be that when the system was installed they pressure tested it with N2,
We had this same issue when moving into a new facility last year. N2 at varying levels dependent on which part of the manifold we plugged into.
Solved by flushing every point of use.

I'm with Peter, dont use snoop for any GC or GC/MS system. I use 50/50 IPA and water.

[MAForensics]

The pressure test with N2 would be an interesting thought, but we have no issue plugging one of our other systems into the same He nozzle. Only the new instrument shows a response at that He outlet.