Poor %RSD with 6890 GC

[stappg]

Hi Guys,
I've got a problem with poor %RSD on both area and height. Instrument is Agilent 6890N with EPC and a 7683 ALS. Analysis is ethanol, volatile impurities as per BP2004. Conditions are: column AT624, 30m x 320um x 1.8um, He, FID, inlet temp=200C, detector temp=280C, split 20:1, velocity=35cm/sec, inject 1 uL, oven prog 0-12=40C, 12-32=40C-240C, 32-42=240C, mode is pulsed/split at 0.2 min and 30 psi. I'm using new 10 uL syringe with teflon tip plunger, new 4711 liner with wool removed, 5m x 320um inert retention gap, new gold seal, septa and vent filter. Data aquisition via Empower. MeOH and acetaldehyde are approx 10 ppm in EtOH, resolution is >1.5 but MeOH peak has very bad tail (maybe second or even third peak but no MS to check??). %RSD for each peak is approx 25%. When injected at 1000 ppm %RSD is approx 10%. Same solution injected on old Shimadzu 17A gives <1.0 %RSD at both concs. I've also tried injecting a solution of EtOH with toluene at approx 4000 ppm and %RSD on toluene is still 7%. I'd expect high %RSD's at low ppm's but why does 17A give acceptable results? Surely the 6890N can do better than this??
Any ideas?
Any suggestions?

Thanks, Peter

[Victor]

1) Do these methods use internal standards? Split injection can be quite irreproducible without them.

2) Are you injecting manually-this is another source of irreproducibility compared with an autosampler.

3) Possibly 1ul is quite a large injection volume even though you are doing split injection. It is possible the injector liner is not able to accommodate the initial volume of sample vapour and that back flushing is occurring.

4) What is AT 624? Some phases used for ethanol analysis solidify around this temperature and give poor results. This is a wild shot though, I'm sure you checked it.

5) Maybe you should try replacing the wool. Wool is inadvisable for highly adsorptive compounds but I don't think you should have problems with the solutes you are using.

[DR]

If you are injecting too much for the liner to handle, you should be able to program the injector to do a very slow injection (works well with nasty flashing solvents like chloroform as long as you cold trap on a retention gap).

[JI2002]

If you are doing split injection, why you have a holding time of 0.2 min? It looks like you are doing splitless injection.

[sg]

When you use X-624 (such as DB-624, HP624, your column, BP-624 etc....), the methanol normaly has very poor tailing, the typical value is about 2.5.

One question is that why you use pulsed split mode for such high concentration solution (1000 or 4000 ppm?). For the low concentration such as 10 ppm, I think maybe your pressure or the time is not sufficient resulting in irrepeatable results.

Why remove the wool in liner? Can try a liner with wool. your solution injected may not vaporize completely.

[stappg]

Thanks everyone for the reply so far.
AT-624 is equiv to USP type G43 (slightly polar).
Method for EtOH (volatile impurities) is from British Pharmacopeia monograph and cannot be altered much (some parameters + or - 10%). No internal Standard is used. Injection is via autosampler.
My thinking was that some activity may be occuring (in liner or wool) thus causing MeOH peak to split, so I removed wool and introduced pressure pulse at start for 0.2 min. Seemed to help but not completely fix??
4711 liner volume is 870 uL. 1 uL of EtOH expands to about 400 uL at 200C and 10 psi. So liner volume shouldn't be a proble? Especially if pressure is 30 psi for first 0.2 min.
Thanks again.
Peter

[gcguy]

You seem to have quite a problem. We analyse for various alcohols and other volatile compounds and a 624 type column is usually our first choice. The Agilent autosampler can give non-reproducible injection volumes, this can be improved slightly by using a 5ul syringe.
Is your column ok? Have you used the column for this method before, was it ok then? You could have a fitting problem such as incorrect injector end insertion length or a bit of ferrule inside the column. I would be tempted to start again with the whole of the instrument set up.

Best of luck.

GCguy

[stappg]

Thanks guys, further to discussion:
I removed approx 1.5 m from each end of the column and I have ordered a 5 uL syringe (should be here today??). The poor %RSD is also evident with other methods, using different columns, although the current 624 is new but appears good (it was conditioned correctly). I also ran a method that used the front inlet set up with a packed column and this gave %RSD's of about 4%. Is the ALS 7683 Injector inherrently bad with respect to poor %RSD? Can I expect to reutinely achieve better than 4% RSD? (our Shimadzu 17A achieves <1 %RSD at 10 ppm).
Any further suggestions???

Thanks again, Peter

[Rick Jagielski]

With our 6890 we typically get %RSD <0.5 for 10 consecutive injections. Usually we only see the injection precision suffering when we have neglected to perform our PM in a timely manner.

I have analyzed for ethanol, methanol, and IPA at moderate concentrations using Sec-butanol as a carrier solvent and gotten good results, (the method competed validation and was used for several years) I believe this was one of the few assays we had where the analyte peaks eluted before the main solvent front, but it gave us good results.

When RSD suffers with our instrument we usually do what you have already done, Replace syringe (we use a 10ul), clean inlet and vent traps, replace seal and liner, trim the ends of the column.

A leaking fitting (usually the inlet side column ferrule) can cause a very noisy run for all injections on that column. If you are having precision problems on both columns/inlets/detectors (front and back) then I think the only place left to look is your autoinjector.

Do you leave some headspace in your vials?

Rick

[WK]

Stappg
I agree with sg.
Try packing the liner with wool and see if your %RSD improves.
I get appalling peak shapes for alcohols without wool.
Maybe you need to pack the liner in a different way or with different wool.
When I have used pre-packed liners for my PE GC I have ended up packing them myself.
I find the pre-packed liners have insufficient wool for my split injections and also its not packed tightly enough.
WK

[Peter Apps]

My first step would be to simplify everything - go back to a straightforward split injection, and lose the retention gap, which does nothing useful if you are injecting split - I suspect that the tailing is due to the retnetion gap and its connection to the column.

The default fast injection on the Agilent autosampler is often not optimal, try a 2 or 3 second pre injection delay to let the needle heat up in the inlet, then a slightly slower injection than the default. This simulates a manual injection and I have found that it sometimes gives good results with solvents that have higher boiling points or a higher heat of evaporation than the usual solvents like pentane, hexane and dichloromethane.

Good luck Peter A

[shashikant]

I think that there should not be much of a RSD problem in HP-6890. Please use a Split mode, if possible increase the split ratio to 25:1. i If area counts are not an issue). Please check first that what is the situation in Splitless mode. It may be the reason that the injector itself is giving some Leak.
Give pre-rinsing 5-6 times before the injection. I presume that the column tip is properly cut and installed into the injector port and leak test done.

I have done lot of work with various solvents using HP-5890 and HP-6890. It gives consistent RSD's. ( I feel it should be beeter than Shimadzu's.)

Hope that serves your purpose.

Rgds
ShashI Kant Tiwari

[wsayers]

You should definitely run this analysis in split mode. 6890s are forepressure regulated, so the "pulsed" mode is there primarily to improve peak shape for difficult compounds in splitless injections.

Pulsed mode runs the injection at a higher than normal pressure to assist in getting the analytes onto the column quickly. After injection the inlet backs the pressure off to whatever is required to maintain your selected column flow rate. This can cause peak shape distortions so don't use it unless you need it.

You might also want to increase the split ratio slightly (say 50:1), and run your flow rate up to 40cm/s. Always use the smallest syringe that will deliver the desired injection volume for best precision.


Best of Luck