Inconsistent response from Agilent 7890/5975C

[wss]

In my lab we run a purge & trap setup using a Tekmar unit plumbed into an Agilent 7890 GC/5975C MSD. A week or two after doing a service on the GC and MSD (replaced liner, septum, column, cleaned source, replaced old filaments, etc.) the response of our instrument-added internal standard started to go all over the place - sometimes doubling between runs, sometimes consistent for a couple of samples then low or high the next.

We initially suspected the Tekmar, but ruled this out as we have exactly the same problem when using a 7683 injector turret to inject standard directly.

I did the first inlet leak check as described here, but the final pressure after the prep run has started is as it should be: https://community.agilent.com/docs/DOC- ... od-changes

The tune looks good, no obvious issues with the MSD. I have noticed that the rotary pump is leaking oil slowly, but this doesn't appear to affect the vacuum at the instrument.

Any suggestions as to what might cause these inconsistent responses?

[benhutcherson]

Are you running split injections?

I didn't see you mention the split trap in your consumable replacement. If not, it's a worthwhile place to check. I've seen inconsistent peak areas from this trap being saturated/partially blocked.

[Jesse Bischof]

Could you go into detail a bit about what inconsistencies you're seeing? Also, what are you analyzing?

[James_Ball]

Does it affect the higher boiling 1,4-Dichlorobenzene-d4 more than the lower boiling Fluorobenzene internal standard(if you are using those)?

Have you checked the mass alignment? Look at the spectra of the internal standard in question and zoom in on the quantitation mass and see if it has shifted(especially to the lower side) from what you have set in the quant database. If it shifts enough that one run you get the whole amount and the next the majority of the peak is just outside the mass window it looks for then you can get this type of problem. Can usually be fixed by simply doing a mass update in the quant software.

[wss]

Are you running split injections?

I didn't see you mention the split trap in your consumable replacement. If not, it's a worthwhile place to check. I've seen inconsistent peak areas from this trap being saturated/partially blocked.

Yes we are running split injections, I don't believe we have a split vent trap installed on this instrument.

Could you go into detail a bit about what inconsistencies you're seeing? Also, what are you analyzing?

We use the instrument to analyse for the USEPA 8260 suite. In terms of inconsistencies, there's almost a doubling and halving effect between runs. Take for example our dibromofluoromethane surrogate - the responses over 8 runs will look like this:

839043.5485
466252.5515
808309.0815
555656.404
460504.9385
750522.6485
445331.6745

Does it affect the higher boiling 1,4-Dichlorobenzene-d4 more than the lower boiling Fluorobenzene internal standard(if you are using those)?

We're not using exactly those standards, but it appears to affect the lower boiling standards more than the higher boiling ones (e.g. much greater swing between runs).

Have you checked the mass alignment? Look at the spectra of the internal standard in question and zoom in on the quantitation mass and see if it has shifted(especially to the lower side) from what you have set in the quant database. If it shifts enough that one run you get the whole amount and the next the majority of the peak is just outside the mass window it looks for then you can get this type of problem. Can usually be fixed by simply doing a mass update in the quant software.

I haven't checked this yet, but I'll give it a look today. Thanks.

[James_Ball]

Are you running split injections?

I didn't see you mention the split trap in your consumable replacement. If not, it's a worthwhile place to check. I've seen inconsistent peak areas from this trap being saturated/partially blocked.

Yes we are running split injections, I don't believe we have a split vent trap installed on this instrument.

Could you go into detail a bit about what inconsistencies you're seeing? Also, what are you analyzing?

We use the instrument to analyse for the USEPA 8260 suite. In terms of inconsistencies, there's almost a doubling and halving effect between runs. Take for example our dibromofluoromethane surrogate - the responses over 8 runs will look like this:

839043.5485
466252.5515
808309.0815
555656.404
460504.9385
750522.6485
445331.6745

Does it affect the higher boiling 1,4-Dichlorobenzene-d4 more than the lower boiling Fluorobenzene internal standard(if you are using those)?

We're not using exactly those standards, but it appears to affect the lower boiling standards more than the higher boiling ones (e.g. much greater swing between runs).

Have you checked the mass alignment? Look at the spectra of the internal standard in question and zoom in on the quantitation mass and see if it has shifted(especially to the lower side) from what you have set in the quant database. If it shifts enough that one run you get the whole amount and the next the majority of the peak is just outside the mass window it looks for then you can get this type of problem. Can usually be fixed by simply doing a mass update in the quant software.

I haven't checked this yet, but I'll give it a look today. Thanks.

If you are using a split/splitless inlet then there is a split vent trap, it is located under the rear top cover of the instrument behind the injection ports. The split vent is the copper line that comes out of the injection port and runs back to the split vent trap, which is an brass/steel fitting that screws apart to hold the carbon trap, then flows to the backpressure controller on the EPC.

We had a Centrurion autosampler doing that a long time ago and we had to replace the Internal Standard valve. Not sure how the Tekmar one works, I have used their system since the old LSC2000/2016 model.

If it affects only certain compounds and not all of them equally then it could be the mass alignment. Also it could be some contaminate in the flow path that is being alternately put into play. The old Tekmar 2000 had a 6 port valve that rotated forward with each sample and if one groove in the rotor became contaminated then every third sample would show it. Not sure if the new Tekmar units still rotate forward each time of do like the Encon and rotate forward then backwards with each sample.

Also put a flow meter on the split vent outlet and watch it during a few samples. If one part of the valve is contaminated or clogged you may see the split flow changing during desorb. We had a purge and trap that the valve motor had a problem and ever few samples it would not completely switch the valve and the flow would drop more than half during desorb which made it have internal responses all over the place.

[wss]

Thanks for the suggestions everyone. The problem has resolved itself, but I'm not sure if I've ascertained the cause... We noticed that the little vacuum pump (a Pfeiffer Duo 2.5) had leaked most of its oil (though it wasn't empty). We replaced the pump with a spare Agilent MS40+ pump from one of our LC triple quads and, though the problem persisted for a bit, the surrogate and internal standard responses became nice and consistent again when using the autoinjector, then again as well with the Tekmar purge and trap.

What was curious is that the response dropped suddenly just before consistency returned. So, we have lesser response, but at least they are consistent. I have to check how much lower they are than when the instrument was last operating optimally.

I've replaced the pump seals on the little Pfeiffer pump and it's become our 'spare' pump until we need to commission a new system.

The Agilent MS40+ seems like overkill for the task. It needs to be operated manually from a power point as the internal switching on the 5975C* can't handle more than 700VA on the RV pump outlet. We just switch it on before we switch the MSD on. Any experiences using these pumps long-term for GC-MSes?

*n.b. the unit in question is actually a 5975C, not a 5973C as stated in the OP, sorry for the typo.

[James_Ball]

I have not used anything as large as the MS40, but all of my 5973 and 5975 are using E2M2 sized pumps. The little ones just run too hot and don't last very long. My larger pumps are plugged directly into the wall.

It could be that the larger pump had managed to pull any residual moisture out of the system. If you have moisture in the system, even if is it downstream of the source which doesn't show up so much in the spectra during tune, it can cause the response to be higher, especially for higher mass analytes. If you scan just after having the MS vented you will notice the 502 peak is larger than it is after 4 hours or even after it pumps down over night. Larger pumps shouldn't hurt and I think actually help keep them cleaner over time.

I believe SIS (Scientific Instrument Service) published a paper on installing a pump similar to the MS40 or slightly smaller and it showed an overall improvement, but that was a long time ago.

[wss]

An update - the MS40+ performed flawlessly for a couple of weeks, then I came into the lab the other day to find that it had shut itself off leaving a little glowing red LED as the only sign of life. I never really liked the idea of pump having any kind of electronics in them! I haven't looked into what the failure mode could be yet, other than maybe overheating protection (though it was in a well ventilated area). It repeated the same problem after a power cycle the next day (and nearly delayed a significant number of samples).

Time to refit the old Pfeiffer pump, which I've replaced the gasket on, to see if it makes a difference.