GC peak variablity

[mbair]

I'm injecting 1 uL of a petrolatum/mineral oil formulation at a solution concentration of 5 mg/mL in cyclohexane. My instrument is an Agilent 7820A. The inlet conditions are 280 C, 2:1 split, and the inlet liner is a typical 4 mm split glass liner with glass wool plug. Column flow is 2.2 mL/min He (~44 cm/s), with a temp. prog. from 50 C to 325 C. There are 4 component peaks of interest - all of them elute prior to the jelly/oil base components, which just get baked off the column. My components of interest are at very low levels in the sample material (ranging from 0.003% to 0.3% w/w), but the method offers good detection and excellent peak shape and resolution.
When injecting "clean" solutions of just my component standards in cyclohexane my standard bracket precision is good, often < 1% RSD. However, after injecting a bracket of my oily/gooey sample solutions, the last peak of interest (least volatile) jumps-up in area by ~4-7%. Only after several "clean" injections does the peak slowly come back down. The other peaks of interest vary only slightly (~1%).
I'm fairly convinced that it is an inlet issue. It appears that injecting the gooey/oily stuff affects the column loading of this one late-eluting peak for several injections to follow.
My questions are: What could be causing the signal response irreproducibility of just this one peak, and how can it be most easily remedied? Increased inlet temp. and/or split flow? Different liner? Different solvent or decreased solution concentration? Making repeated "blank cleansing" injections? What do you suggest?
Thanks! - MDB

[James_Ball]

I would start with increasing the inlet temperature, since it could be allowing the heavier matrix to hang around in the inlet if it isn't volatilizing completely or rapidly enough. As long as your target analytes are not broken down by the extra heat you should be ok.

[Peter Apps]

A 2:1 split does not give enough flow through the inlet for the pressure control to work effectively, and in addition the flow is too small to flush the heavy compounds out between injections. You would probably do better with a splitless injection of a more dilute solution, and then set a high split ratio (say 50:1) so that the inlet is thoroughly purged.

Peter

[mbair]

Thank you for your insights, gentlemen. You've helped me confirm what I was initially suspecting. Today I'm trying the following:

Increase inlet temp from 280 C to 300 C.
Decrease split flow from 2:1 to 1:1 (splitless)
Induce purge flow @ 100 mL/min after 0.5 min for 2.5 minutes duration
Halve injection volume from 1.0 uL to 0.5 uL.
Double sample conc. from 5 mg/mL to 10 mg/mL.

Single injections under these conditions look good so far. Waiting to see how the repeatability rates. Thanks again! - MDB

[Peter Apps]

A split ratio of 1:1 does not give a splitless injection, it will make the inlet flow even more difficult for the flow controllers to control. Set the inlet to proper splitless with the split purge opening after 30s. Depending on the syringe volume 0.5 ul will give poorer injection repeatability than 1 ul, so rather leave the injection volume and the concentration the same, or leave the volume the same and halve the concentration.

Peter

[mbair]

It looks like the issue is persisting. I still get a gradual increase in signal response with the splitless injection mode + split vent purge.
I will note that, when I began developing this method, the precision was great (RSD < 1%), and now I'm lucky it it is 2-3%. The chromatography still looks great - good peak shape and efficiency, and no baseline issues. It just seems like the precision is getting slowly worse with each injection.
Is there anything else that I can try? Is it possibly that the high-boiling material is building-up in the column? Should I try a solvent flush of the column? Try changing some of the inlet hardware (gold seal, etc.)? Please advise.

Thanks.
MDB

[James_Ball]

Looking up the boiling point of petroleum jelly it lists it at 343C which is above your highest column temperature so you could be seeing a buildup of some of the high boiling fraction on the column over time. If your column is compatible with solvent flushing you may be able to reverse flush it with solvent to clean it. If you are not using a guard column you may want to, since it can be clipped every few days without changing the retention times much and would eliminate some of the high boiling contamination if present. You may also need to rinse out the injection port for the same reason with it being so near the boiling point.

Another thing to do is rinse out the split vent line. On the Agilent split/splitless injection ports it is the copper line attached with the brass swagelok fitting at the side of the port and another near the rear of the GC. We have found that you need to remove this portion of line and solvent flush it since high boilers will condense in that line with it being near room temperature. This can cause flow problems and breakdown issues believe it or not. I have gotten in the habit of just replacing that portion of copper tubing with fresh tubing, just cut to length and bend it to match the one already there.

As for having 2-3% RSD injection to injection I would love to have that good of repeatability, but using a purge and trap that is very difficult to hold over very many injections

[Peter Apps]

How frequently (number of samples) do you change the inlet liner ?, and when you do change it is there any visible muck in it ?

Peter

[Yama001]

It may be useful to follow an analytical run with one or more solvent injections with the system at its max temperatures and at a high split. This will help clear the inlet of some of the gunk. I do this routinely for fuel fingerprinting.

[mbair]

Thank you all for the helpful replies. To update you all, I have recently done the following:

Pulled out the S/SL inlet port and split purge line to wash with solvent (there was no visible gunk build-up present).
Installed a new liner (I have recently begun using splitless injection with a 1.5mm id strait liner, along with a 3 min split vent purge @ 100 mL/min after 0.2 min in an attempt to keep the inlet clean)
Installed a different column (temporarily replacing the DB-1 column with a HP-5 of similar dimensions. I attempted a solvent flush of the old column, but so far no luck).

As a result of these actions, I have seen some improved precision, but like before, my peak areas begin to creep-up by ~3-4% once I start injecting my greasy samples. This drift upwards typically happens after injecting my first samples, so after repeated sample/standard brackets, the values tend to level-off and my precision returns to < 1% RSD. As usual, this result isn't consistent for all of my peaks of interest - almost always worse for the one last eluting phenolic compound, and often at least one other peak as well, but that can vary. Regular blank injections show no carryover and only modestly improve precision. Chromatographically speaking, the baseline, RT, and efficiency all look very good and are very consistent.

It certainly appears that the high-boiling components are building-up somewhere in the inlet or the column. I currently suspect either: 1) the liner is getting coated with grease, blocking Si-OH adsorption sites for these compounds allowing more to be volatilized, 2) the split vent trap is getting gunked-up, causing loading issues (however, I would expect more random error and less drift with this cause, and a dirtier transfer line), or 3) build-up of grease at the column inlet is causing preferential partitioning of the analytes onto the column.

I believe this issue is a mix between #1 and #3. I'd had better luck with the 4 mm liners w/ glass wool - they seem to work better for a longer period before this peak area drift takes effect. My next action is to try a (pulsed?) splitless with these liners w/ split purge and do my best to keep these liners clean and/or replace regularly. These glass wool liners - can they be solvent washed (e.g. w/ DCM?). For $30 a piece, I want more than just a dozen injections from them! Thanks again.

MDB

[Peter Apps]

For the splitless injection that you are now doing a 1.5 mm i.d. liner has too small a volume to accommodate the cloud of evaporated sample. You will be better to go back to the 4 mm liners.

Your splitless time of 0.2 min is on the short side, I would start with 0.5 min. Also, do not turn the split flow down again after 3 min, leave it at 100 ml/min throughout the run. This sounds extravagant in gas, but look at it as a troubleshooting step, you may be able to reduce it later.

Just so that we do not get too focussed on the inlet - what detector are you using ?

Peter

[mbair]

Peter - Thanks for your replies.

I am running FID at 320 C. My inlet is at 290 C, which is just hot enough to ensure fast vaporization of my least volatile compound of interest (BP = 265 C). My max column temp is at 310 C (ramped up form 50 C). I also ramp-up the column flow to help push-out the goop. Typically, I like to have a greater range of temps, but for this application with high-boiling components, I'm pushing the upper limit of my operating range.

I had been using 0.5 uL injections in the 1.5 mm splitless liner - all calculations for cyclohexane at 290 C and inlet pressure at ~20 psi indicate that I am at ~65% volume capacity for the liner, and there was never any indication of backflush (random precision of a clean standard gave routinely gave RSD < 1%).

Right now, I've gone back to the 4 mm liner w/ wool. Also, using pulsed splitless mode, I've found that 0.5 min purge time with this inlet configuration gives me optimal column loading. My idea is to keep the high-boiling components trapped in the wool throughout the course of a typical run, then do my best to sweep them out afterward. I start seeing this peak area increase only after the wool gets saturated with gunk - that is my current guess, at least. It explains why this effect tends to get worse with time.

Just as a side note - I've moved this method over to GC only after attempts at HPLC quantitation proved more difficult (multiple extraction sample workup = timely and imprecise, poor solution stability for one analyte with water, residual ointment jelly crashing-out in the MP, higher detection limits, etc.). Despite all the issues with the GC method, I still consider it easier and more reliable than any iteration of the RPLC method that I've run. I haven't tried normal phase LC yet - that's a last resort!

MDB

[AmericanSciPro]

These glass wool liners - can they be solvent washed (e.g. w/ DCM?). For $30 a piece, I want more than just a dozen injections from them! Thanks again.
MDB

Any repeatable expense can be a burden, I get that. What cost would make it worth it to you to not clean them? I’ve considering producing simple liners like this and just sort of having a rough ballpark would help in figuring if it is worth the effort. Thanks…