Headspace Unretained Extra Peak

[sdegrace]

I have a weird problem that I wonder if anyone has encountered before or can shed some light on.

I have an Agilent 6890 GC with an Agilent G1888A headspace sampler. Recently I changed the G1888 from manual pressure control (MPC) to electronic pressure control (EPC), for the carrier only. That involves cutting the carrier line and redirecting the flow through the HSS so that the only carrier source for the inlet is through the HSS transfer line. I connected the transfer line close to the inlet and wrapped the exposed section of tubing with glass wool and aluminum foil to minimize any cool zone going into the inlet. I don't use an autosampler tower on that inlet.

There is one method which heavily dominates the use of the HSS right now where an inorganic salt with a highly oxidized organic ligand is digested in 44% w/w phosphoric acid and the solution in phosphoric acid analyzed by headspace GC. This method continues to work well with no surprizes after the change.

However, I started working on a new method using DMF as the diluent, and I started seeing something weird, which I can't remember ever seeing before I changed the instrument.

Simple background on the method - uses a 60m x 0.54mm x 1um Phenomenex ZB-WAXplus column and starts out on a gantle ramp from 40C to 60C over ten minutes. The linear velocity is held constant at 20 cm/sec. Right now the HSS zone temperatures are oven 135C, loop 155C and transfer line 165C, but the same thing is seen at lower temperatures. Inlet temperature is 180C.

When I first inject a vial of DMF (20 mL vial, 1 mL in the vial), I get a monstrous peak at just slightly later than five minutes. This peak tails badly even though no other peaks are showing signs of tailing except TEA (even the DMF peak looks pretty good). Note that for a 60 m column at 20 cm/sec, this peak is pretty much unretained. With successive injections of DMF, the peak becomes smaller and smaller until it reaches a quite manageble size which doesn't interfere with anything else I'm looking for. It always tails. Even after leaving the instrument idle for a while the area never goes up again, as long as the instrument isn't used for other method in the meantime.

If it's a real thing I figure it would almost have to be a light hydrocarbon, because even heptanes are retained on the column, and anything even slightly polar would probably be quite well retained on that wax. That doesn't make any sense at all to me, though.

I put my work down to go on vacation and many more runs of the method with the phosphoric acid diluent we done. When I came back I took the HSS back again to work on my method with the DMF diluent, and lo and behold, the problem is back, with the exact same signature, a giant apparently unretained peak rapidly dimishing as injections go on. Swapping out the inlet liner has no effect. This is a split method, I'm using a Phenomenex split/splitless liner with glass wool.

Does this ring any bells for anyone? I'm also pursuing it with Agilent to see if they have any ideas.

Thanks!

Stephen

[krickos]

Hi Stephen

This rings a couple of bells.

Most likely this peak is related to the high HS oven temp of 135°C. I have seen traces of this peak at 70°C and at 100°C it becomes as you describe ie always tailing and in my case now and interfering with methanol peaks on DB-624 columns.

Why is it here? Well not 100% sure but it is either volatile impurities in DMF or impurities that are formed under prolonged exposure to that temperature and pressure.

In some previous investigation (WAX column) we have identified stuff like dimethylamin and methanetiol, things that would appear in the front or close to. In some cases this peaks also have been more common in basic sample solutions (water in sample seems to speed up formation).

Simplest way to confirm this is to lower HS oven temperature to lets say 60-70°C and see if it is reduced/eliminated.

[sdegrace]

Well, I've tried injecting DMSO but I still see the same peak - I'm developing the method myself so I have full latitude, and if the DMSO worked better I could just switch to it. But I'm still seeing the peak. That isn't to say that the DMF is not the ultimate source of the problem by introducing the contaminant somewhere in the system, maybe. Agilent recommended I try following their steam cleaning SOP so I'm going to give that a whirl. I'll let you know how it turns out

Stephen

[sdegrace]

Also re temperature, I was using 100C oven temperature before when I had the same problem but I didn't try reducing the temperature further, I'll give that a go on Monday after I run the cleaning, if the cleaning doesn't resolve the issue for me, and I'll let you know how that goes.

Thanks!

Stephen

[sdegrace]

I performed the steam cleaning Agilent recommended and then injected a sequence with my method as follows:

Air Blank
DMF
DMF
DMF
Air Blank
Air Blank

(1 mL DMF, using a 20 mL HSS vial from Agilent).

All of the injections including the initial air blank had the large, tailing, poorly-retained peak, with a general decreasing trend, although not always decreasing from one injection to the next (e.g., the peak increased somewhat between the initial air blank and the first DMF injection).

I'm feeling that the problem is not in the samples at all, which leaves me a little stumped. I also really don't believe the problem is the 6890 - I really feel in my gut the trouble is coming from the G1888 somewhere (maybe related to my change of the system over to EPC? I don't see how, though). I wonder if I should purchase a PM kit and start replacing some parts, although the instrument is not due for a PM for another couple of months. I would love to pinpoint the cause of the issue, though.

Stephen

[Peter Apps]

Hi Stephen

Was the extra line that connects the EPC to the headspacer made from cleaned tubing ? A broad hump in the first run after a cleaning suggests strongly that a fairly volatile contaminant is coming in with the carrier gas.

Connecting the transfer line into the inlet carrier gas feed inevitably introduces a cool spot, and exposes the sample to the undeactivated metal of the carrier gas feed. The only reason to do it is if you need to do conventional injections to the same inlet without disconnecting the headspacer, but even then you compromise your carrier gas purity for the conventional injections by passing it through the headspace plumbing.

What do you see if you just run a programme blank on the GC - i.e. no injection at all from the G1888 ?

The glass wool in the inlet has no use in a headpace injection - try removing it.

Peter

[sdegrace]

Hi Peter,

When I changed the HSS to EPC, I purchased a kit from Agilent containing all the unions and tubing I would need. The carrier connects via zero dead volume union that came with the kit to a piece of tubing flared at the end to attach to the Swagelok fitting on the HSS that also came with the kit.

To try and avoid creating too much of a cool spot, I made the bare tubing from the transfer line to the inlet as short as possible, then wrapped it with glass wool and aluminum foil.

When I run a GC program with no injection, I see the normal, expected profile with no additional peaks.

I tried running some HSS injects with air blanks and DMF but with much reduced zone temperatures (oven 70C, loop 105C, transfer line 115C) and indeed my peak up front is smaller.

I think I certainly could have compromised my carrier gas purity, but since I only see the phenomenon when the carrier passes through the loop (i.e., on HSS injection), it makes me think the source has to be in the sampling part of the G1888, especially since I also see it when injecting empty vials.

Right now, I'm thinking strongly along the lines of doing a little PM on the HSS. I was thinking of replacing the probe, the loop, and the deactivated tubing. I'm still curious as to what the heck is going on, but on the other hand, if I can make the problem just go away I would settle for that.

Stephen

[Peter Apps]

Hi Stephen

One more possibility is a disruption in flow, but I would not then expect the peak to get smaller in subsequent injections, which is a classic symptom of contamination building up while the machine is idle, and then getting flushed off bit by bit with each injection.

When the G1888 is not being used, what are the zone temperatures and flows ? If I recall you can adjust the needle purge flow ?.

Is there any way that you can heat the junction between transfer line and inlet feedline - a hot air blower or a soldering iron or somesuch, initally just to see if it is the seat of the problem ?

Peter

[sdegrace]

Hi Peter,

When the HSS is not in use, it tends to get left on, so the zone temperatures will depend on the last method run. There is one method which is heavily dominant right now and it uses temperatures of 70C for oven, 105C for loop and 115C for transfer line.

As to flows, that is an interesting point. Before when we had MPC, either we had the whole instrument turned off when not in use to conserve helium, or else there was a constant flow of whatever was set manually on the G1888. Now, it depends on whether a column is connected. If there is a column connected there will be a constant flow of some kind, but if changing the column, or changing the liner, the flow will be off.

Maybe I'm creating problems by having heated zones while potentially there is no positive pressure of inert gas to protect the parts from oxygen?

At least according to what's written in the G1888 manual, there are only two pressures I can adjust, the vial pressure and the carrier pressure, so I think my options may be somewhat limited.

I'll try to think of some way to heat that little zone of tubing going into the inlet... I do have a heat gun which might do the job. I hope that's not it because I don't want to reconfigure the instrument once again and I like the advantages of the current configuration very much. Not having to manually measure flows and calculate the real split ratio simplifies things considerably.

The fact that I still get the contaminant peak even when doing an air blank makes me think it's probably not that connection, though, because in that case I should not be passing any vapour through that part which could dislodge anything.

Stephen

[AICMM]

Stephan,

I am going to take the opposite tack of Peter here. Unless I am missing something, you say the peak looks like an unretained peak. If unretained, it is probably not line contamination since that would increase your signal all the time, not come out as a peak. So if it appears more like a peak, you need to look at things that act like injections. Valves turning on and off immediately come to mind. As you do it more and more, you allow less time for the contamination to build up at that spot....

If you were going to troubleshoot this approach, you leave the oven fairly hot and you turn valves on and off sequentially and wait long enough for something to show up at the detector.

Just a thought.

Best regards.

[Peter Apps]

Hi Stephen

The difference in temperature between the two methods is a clue. I think that there is something building up while the phosphoric acid samples are running. You then increase the temperatures, which begins to displace it. That you have a peak rather than (or maybe in addition to) a constant background signal means that the pulse of flow, or sample vapour that follows an injection displaces more of it.

Changing bits on the G1888 might help, but then again it might not.

Can the phosphoric acid samples be run with the loop and transfer line temperatures that you use ? That might stop the contamination building up in the first place. Probably not the first choice if the method would have to be revalidated.

Would it be pratical to run a few clean out blanks in your sequences before you run samples ?

My money is still on that cool spot. The alternative route for the transfer line is in through the top of the inlet. Agilent will sell you a needle to go on the end of the transfer line ( for a remarkably high price !) which you stick through the septum into the hot zone of the inlet. Or your could go for the volatiles interface (which is a fancy four-way union) that you mount as a second inlet. Some heater tape around the cool spot is likely to be cheaper.

Peter

[sdegrace]

Hi Peter,

It would make me very sad to have to go back to the MPC setup.

We used that type of approach before last month on the G1888. The problem I have with it is that you have two sources of carrier gas, the HSS transfer line and the carrier line from the GC's EPC module. The HSS carrier flow is adjusted very roughly with a knob (it's quite cranky and unpleasant, at least on our unit, actually), so you're always measuring the flow coming out the split vent to figure out if you have enough flow from the HSS, and your split ratio is not the one you set in the method, you have to calculate it.

I wanted to change to the single-source EPC setup for the simplicity and repeatability it confers to the methods (less need to train people, less scope for human error), which makes the robustness portion of method validations easier because you do not need to look at such a wide range of HSS carrier flows and split ratios, and the method itself doesn't have to be quite as robust in these areas.

I was a little freaked out when I reviewed the Agilent documentation and discovered the setup they proposed with the cool spot, but it was Agilent themselves advocating this approach, and that made me figure it must actually work in practice. That said, I took steps to try and mitigate the cool spot, by making the tubing short and insulating it.

I will be most displeased with Agilent if it turns out that fundamentally some really common methods e.g. residual solvents in DMF, just don't work with the setup they themselves recommend and they didn't warn me about it.

One thing I could try, maybe, though, is to leave the EPC flows as-is, but to connect the transfer line through the septum and needle. I have a couple of needles around (yes they are not cheap!!) from the previous setup. Presumably I could purchase a plug to screw into the zero dead volume union that is connecting the transfer line to the carrier gas line right now. That would create a little cool appendix to the inlet, though, and I'd worry about that, too.

I think I'm going to wait for the PM kit and do the PM first, because it's not far from its scheduled PM anyway, and who knows, that might fix the problem. I'll let you know how that goes. Unfortunately, I can't touch that method with the phosphoric acid, that would be a whole can of worms. But that method has been running for close to a year now and this is the first time other mehtods have been giving problems, so it's difficult for me to accept that I should have to change it anyway.

Stephen

[aldehyde]

There are a lot of potential causes, do you have filters on your gas line? Have you changed tanks recently? Are the tanks near the instrument or are there lines coming out of the wall that extend for god knows how far through the building?

It is possible that you have contamination of some part of the pneumatic path, a damaged or malfunctioning valve (vial pressurization and vent valves get hit with sample during every injection and they commonly fail). One way to test for proper valve operation is to connect a piece of tubing to the vent exit on the back of the headspace where you hooked up your carrier line. Put it in a vial of water, go to the advanced functions > 5 (manual operation--change the parameters with the 1 and 0 buttons), and then with a crimped empty vial on the probe open the vent valve, you should see a stream of bubbles. Turn off the pressurization valve and it should come to a stop. Turn the sample valve on and repeat this, never-ending bubbles or no bubbles are bad signs. You should be able to hear the valves actuate when you move them.

You can also use this to "manually" inject and try adjusting temperatures, pressures etc to see if you can effect the size and shape of the peak.


I agree that the cold spot is probably not good and could definitely be the source, but troubleshooting this kind of problem is so annoying due to how long the pathway is. So many possibilities.

[Peter Apps]

Hi Stephen

Plugging the tag end of the inlet line and using a needle that you already have should be fairly quick and easy, and simple to undo if it does not help. There will be a bit of a dead volume, but it is behind the needle tip in terms of gas flow, and anything that goes in that direction should be dealt with by the septum purge.

Tapping headspacers into the carrier gas line goes way back to the large volume flows and wide peaks of packed colunm days and GCs with only a single inlet that had to remain available for other injections. I just cannot see the sense of having a heated neeedle, valve and transfer line in the headspacer, and a heated inlet on the GC, probably all with a deactivation treatment, connected together with an unheated section of bare stainless steel. If you can find a thermocouple somewhere put it on the carrier line and union, under the insulation.

Even if you use a needle in through the top of the inlet you still need to reduce the cold spot between the end of the transfer line heater jacket and the top of the inlet - wrap several layers of auminium foil around it, overlapping widely onto both jacket and inlet nut, then put insulation over the foil.

Peter

[sdegrace]

OK, I'll try connecting the transfer line through the needle and plugging off the connection through the carrier line the first chance I get to go back on the instrument. I have some time before the PM kit arrives, anyway. If I can trace my problem back to the cool space, though, I shall have some harsh words for Agilent.

Stephen