Bizarre GC-MS repeatability problem

[100000eV]

Hi All,

I am experiencing a very bizarre repeatability problem with my GC-MS/MS that has so far stumped a few Shimadzu engineers and am wondering if anyone has seen this. The problem is best visualized by the following chart showing the peak area of a methanol solution injected repeatedly on the instrument.



Here are my method and instrument conditions:

Shimadzu TQ-8040nx (GC-MS/MS) with HS-20 headspace autosampler
Column is SH-RXI-624Sil MS 30m x 0.25 mm ID x 1.4 um
HS equilibration temperature is 60 C
Transfer line temperature is 150 C
Sample line temperature is 150 C
HS pressuring gas pressure is 50 kPa
HS vial equilibration time is 22 min
Pressurizing time is 2 min
Split ratio is 50
Column flow 1.9 mL/min
Oven temp is constant at 50 C
Ion source is EI heated to 200 C
MS interface temp is 230 C
m/z monitored is 31 using Q3 SIM

Test sample is 1000 ppm methanol in water
1 mL is placed into a 20 mL crimp-top headspace vial
Peak shape is very good and retention time is consistent
The above data is NOT repeat injections from the same vial, it is 20 separate vials with the methanol solution in each.

I am baffled by the pattern of peak area steadily decreasing in the above chart coupled with the extremely regular up and down pattern. The methanol is being used only as a test standard for repeatability which mimics what the engineers use to test the system repeatability. The method I normally run on this system is terpene and solvent analysis by full evaporation headspace technique. This is not a one-off occurrence. I have performed a weeks worth of repeat injections, column conditioning, etc since the issue started occurring and there is no improvement. This is not a new configuration of this instrument - it has been working flawlessly for the last two years.

Here are a couple of possible clues:

My system is running a twin column. It is configured with the above mentioned column which connects the HS-20 sampler to the MS. The other column is a RXI-5MS, 30 m x 0.25 mm x 0.25 um, and connects the liquid injector to the MS. The injection repeatability of the liquid injector is GOOD - it DOES NOT exhibit this decreasing sawtooth pattern. This indicates to me that the issue is within the HS autosampler and not the detector.
Both columns have been replaced and the issue remains.
This problem manifested directly after the periodic preventative maintenance was performed by Shimadzu. Two years ago, after the annual PPM, this SAME PROBLEM appeared and mysteriously went away after 7 weeks, with no cause determined. For the two years in between the system has been working flawlessly with peak areas consistent across many hundreds of injections and several weeks at a time before needing recalibration.

Any help is greatly appreciated!!!

For good measure, here is the same injection repeatability test performed prior to installation of the new columns showing 40 + injections:



Thanks,
Tom

[70 eV]

You're running SIM at 1 part per thousand and none of the engineers looked at you like a crazy person? I hope you're running that multiplier super cold. You're hammering that column and detector. We run a 120 split on a single quad to hit 0.1 ppb limits in scan mode. Stop throwing so much stuff at your instrument.

[100000eV]

You're running SIM at 1 part per thousand and none of the engineers looked at you like a crazy person? I hope you're running that multiplier super cold. You're hammering that column and detector. We run a 120 split on a single quad to hit 0.1 ppb limits in scan mode. Stop throwing so much stuff at your instrument.

Thanks for your response. I am relatively new to chromatography but try to make sure I am operating the instrument with best practices.

The sample is 1 part per thousand methanol in water. Assuming the water/air partition coefficient for methanol is at least 100, and given our sampling loop is 1 mL and split of 50, plus the 1 mL of sample in a 20 mL headspace vial, there is about 0.01 ug of methanol being injected onto the column with this method. Please doublecheck my calculations and let me know if I'm missing anything. I'm not seeing any saturation of the detector. The peak area counts for the methanol SIM are around 13,000 during these tests. I don't normally see saturation until the peak counts surpass several million.

[James_Ball]

This happens every time after taking the instrument down for maintenance?

Was there any rough pump oil changes in between those PPMs?

If not, it could be the large amount of moisture that is being transferred through the system, loading up into the pump oil and after a few weeks it will come to equilibrium and that is when your areas become stable.

If it becomes stable and then you let it set for a few weeks without running any samples, does it so a slight drop the first few injections before becoming stable? If so that is the same principal just less exaggerated than a fresh oil change.

[100000eV]

This happens every time after taking the instrument down for maintenance?

Was there any rough pump oil changes in between those PPMs?

If not, it could be the large amount of moisture that is being transferred through the system, loading up into the pump oil and after a few weeks it will come to equilibrium and that is when your areas become stable.

If it becomes stable and then you let it set for a few weeks without running any samples, does it so a slight drop the first few injections before becoming stable? If so that is the same principal just less exaggerated than a fresh oil change.

This has occurred after 2/3 previously performed annual maintenances. There were not any oil changes between the yearly maintenance.

It's only ever not in use for 3 days at a time maximum, and I don't really see any difference in performance for the first few injections. I always start runs with 2 blanks to equilibrate things.

I do check the background water using the leak check function of the MS and haven't seen any significant levels of water.

I feel like you may be on to something with the water buildup though. When I first saw this problem manifest 2 years ago I realized that I was keeping the column oven at room temp when not in use. I think I read somewhere that column oven should be slightly heated when not in use otherwise moisture could build up. I switched to keeping it at 50 C when not in use and have been doing so for the past 2 years. Like I said, the GC is rarely inactive other than over the weekend since we are a high volume lab. Of course, I keep helium or nitrogen makeup gas flowing through at all times and it is ultra high purity.

[Peter Apps]

Injecting water vapour would be expected to cause a steady downward trend until everything was in equilibrium.

The sawtooth pattern is more of a puzzle, is there a consistent time between injections?, with some combinations of run time and equilibration time the sampler can be forced to wait for the GC to be ready before it can inject on a regular cycle.

Peter

[James_Ball]

For standby in a volatiles instrument we always leave the oven at 150C to make sure there is no condensation of either water or light hydrocarbons that can contaminate the carrier gasses.

Some of our older instrument actually have moisture traps with molecular sieve in them attached to the rough pump before the line from the MS. I never understood why the stopped supplying these, they helped a lot with Volatiles analysis using purge and trap especially.

[100000eV]

The runtime is 9 minutes but the vial oven equilibration time is 22 minutes. It appears that the system is in "ready" mode when the vial is finished equilibrating, so I don't think there are any issues there.

It's looking like things are improving so I will perform another repeatability test on Monday. It's been about 400 injections since the preventative maintenance and the issue first started.