Agilent 7890/5975C MSD early eluting linearity problems-VOC

[af66]

I've just taken over a 4 year old Agilent 7890/5975C GC/MS system that has not been used in 3 months. When it was last used, things looked normal from a linearity standpoint. Now, it seems that the early eluting VOCs (from Dichlorodifluoromethane to near MTBE) are very erratic in their response when running a calibration range. We normally analyze a 6 point calibration from 10-500ug/ml (we do validation work and the VOCs are direct injected right onto the column) and what I've been seeing is that 4 or 5 of the calibration points seem linear but the 5th and/or 6th point in the series shows the early eluters with a response dropoff of about 70%. I would suspect a leak, but the column program is isothermal at that point and once the responses become normal again, the column temperature has only increased by about 10 degrees. I've put a new liner in (straight taper-glass wool in the middle, similar to what was in there), new merlin microseal, clipped about a foot off the column, put in a new inlet seal, new calibration standards to ensure the integrity of the mixes were intact. I also ran a series of 5 calibration standards from the same vial to test stability with the same vial and the 3rd run in, same issue. I'm focusing on something around the injection port. No leaks present on the MSD end via the air/water check. Kind of stumped at this point. Any advice would be appreciated. As a side note, I have not changed any GC oven or inlet parameters from what had been working prior to this. Any modifications have been strictly limited to consumables.

[Peter Apps]

You are probably seeing detector non-linearity and column overloading. Your top standard is 500 ng/ul and if you are putting that "right onto the column" then it is a huge quantity. Are you seeing flat tops on the peaks, or the sharks fin shape that overloading causes ?

Your description of your injections as "direct injected right onto the column" does not match your use of glass wool in the inlet; I suspect that you are doing splitless injections rather than direct on-column. Direct on-column requires the needle to be in the end of the column, and you need a special inlet to achieve it. With the concentrations you are injecting you need a split ratio of at least 10:1.

Peter

[Bigbear]

Is your solvent methanol? Methanol can affect the elution times for the compounds in question.
As earlier stated you are likely overloading the column, try a split injection of 60-100:1 and see if that improves linearity.

[af66]

Thank you for your responses so far...

To clarify, the method being used is a specialized "in-lab" derived method for direct inject analysis of VOCs. No purge and trap is used. The sample is placed in a 200ul glass vial insert within a 2 ml autosampler vial, loaded on the Agilent autosampler and analyzed much like a semi-volatile sample would via EPA8270.
Admittedly, the calibration range on the high end is excessive but it has been used for 10+ years in here with excellent results. We need the high range to minimize dilutions for this method. We are asked to analyze VOCs over a very wide range of concentrations so it's only to minimize the amount of dilutions needed. The problems now being seen with the linearity issues, appear to be exclusive to the mid range calibration runs (50ug/ml and 100ug/ml it seems).
We have been using a 100:1 split with this method since the beginning to eliminate any effects of column overload, so I think the split is good. Column flow is 1.0ml/min and there seems to be now drop off in pressure from what I can see that would cause an inlet pressure shutdown.

[Peter Apps]

It would be really, really helpful if you could describe the problem consistently and accurately; in your first post you say "We normally analyze a 6 point calibration from 10-500ug/ml (we do validation work and the VOCs are direct injected right onto the column) and what I've been seeing is that 4 or 5 of the calibration points seem linear but the 5th and/or 6th point in the series shows the early eluters with a response drop off of about 70%." in other words the top (or bottom ??) two points are lower than expected. In your last post you say; " The problems now being seen with the linearity issues, appear to be exclusive to the mid range calibration runs (50ug/ml and 100ug/ml it seems)."

Which is it ?; the top end being low points to detector overload, the bottom end being low is probably absoption, mid range with "issues" could be all sorts of things. What exactly are the "issues". We operate under Rod's law (named after a venerable member); the more you don't tell us the more we can't help you.

If your inlet cannot hold pressure you have a leak - whether this accounts for the poor linearity will only be shown by fixing it. From your first post, whether or not a leak is the problem has nothing to do with whether the column is being temperature programmed or whether it is at an isothermal stage in a programme. I recall a similar problem (in real life, not on here) in which an analyst was happily analysing 10% solutions of essential oils with the split set at 10:1. There was a massive leak that was giving him a split of more than 100:1.

And to be clear on the terminology; you are doing a split injection, not a direct injection.

[AaronAIT]

What kind of EM voltage are you running at?

[af66]

Peter, I apologize for the confusion in terminology..In our lab, we run volatiles both via Purge and Trap and also without Purge and Trap when they are injected directly onto the column. We specify the terminology "direct injection" as opposed to "purge and trap analysis" to distinguish between the 2 types of instrumentation being used since both instruments are splitting the sample. I guess I was just on autopilot in describing the method of analysis on the instrument giving us problems. I am familiar with the "split injection" terminology, our lab just tends to distinguish the 2 instruments in that way, hence my "direct inject" comment.
The dropoff in response is in the mid range calibration points. 4 or 5 of the calibration points are linear but either 1 or 2 of the mid-range points (it seems to vary from calibration to calibration as to whether 5 are linear or 4 are linear) drop off significantly. From what I can tell from the GC display, no pressure is lost during those runs that are not linear, however I tend to agree on the thought of a leak, albeit one that seems to correct itself quickly enough where only the first few early eluters are affected.
The confusing part is that when the instrument was last used, there did not seem to be an issue with linearity as what is being seen now. 3 months of idle time and lack of use can certainly bring about changes within the system but there is nothing obvious that points in one certain direction. As far as more details, what I've given the forum is all I've been able to find out at this point...instrument is not linear in the early part of the run (say from when the MS starts scanning to about 3 minutes of a 13 minute run) but perfectly linear from that point on. The split is 100:1 and looking back at the data from the past months, I'm seeing the same area counts in my internal standards and analytes in the calibration runs not affected by this problem as what was being seen back then, so I think the split is accurate. Column was clipped about 12", new microseal installed (no new nut, just the seal), single taper liner with glass wool installed halfway up the liner installed, new gold seal installed. I've been focusing around the injection port, and AaronAIT I had thought about the EM as its voltage is at 950 right now which shouldn't be too bad, but I can't discount an issue with the EM either. I'd have to replace it and check the connections around it to see if there is an issue. I suppose that will be my next step if nothing manifests itself around the injection port.

[Peter Apps]

We need to know all the analytical conditions; temperatures, flows etc and the dimensions of the column.

Please also post an example of a calibration with the problem - there are instructions about how to post graphics as a sticky somewhere.

Intermittent leaks during a run do not affect peak size - all the analytes are on the column within a few seconds of injection, and once they are on the column the only place they can go is to the detector. Intermittent leaks during injection are a different story and they nearly always happen while the syringe needle is through the septum. With back-pressure regulated inlets (which they all are these days) leaks do not affect inlet pressure, so watching the pressure during a run will not tell you anything.

Peter

[James_Ball]

Unless that EM is several years old, 950 is great. That is about where all our new ones start out.

If your standards are in methanol, the time range you are describing is about where the solvent peak would be from methanol. methanol co-eluting with the targets can do strange things in volatiles analysis since you normally don't scan low enough to pick up the methanol, but it is there and trying to be filtered out by the quads while those analytes are eluting.

If you are scanning masses of 44 or below, look to see if you have a higher 44 (CO2) background in the beginning of the chromatogram than after the temperature ramp starts. If so, it is possible that the ferrule on the MS inlet is just a little loose, and as the oven temp rises it seals off. The leak will be so small that you won't even notice it in the ion gauge reading, but it can make a difference in sensitivity of those low mass compounds. It usually only takes about 1/8 turn of the nut to seal it if it is leaking.

[af66]

That is good advice. I'll check that on Monday and see if that is indeed the case.
For reference, here are the parameters currently being used on that instrument...

Inlet Temperature: 210 degrees
Inlet Pressure: 20.221 psi at analysis start
37.3 psi at analysis termination
Total Flow: 104 ml/min yielding 1.0 ml/min after 100:1 split
Septum Purge Flow: 3 ml/min

Oven parameters: 40 degrees initial start time holding for 2 minutes
18 degrees/min ramp to 195 degrees
40 degrees min ramp to 230 degrees then hold at 230 for 5
minutes
Transfer Line temperature: 230 degrees
Column: Agilent DB-VRX 20m .18mm ID 1um film thickness.

Still working on getting a snapshot to post.