Varian versus Perkin Elmer

[aldehyde]

You're right that the best configuration for either a headspace or purge and trap system is a VI (volatiles interface) inlet. Most people wind up adding a headspace to an existing system that already has S/SL inlets or they plan to use the instrument for headspace and non-headspace applications (although even in this case just get 1 VI and 1 S/SL inlet problem solved ).

[Peter Apps]

In my experience, with an ancient Dani, a Varian Genesys, and an Agilent G1888 (which is a Dani clone) the best connection from headpsacer to column is a fused silica transfer line pressfitted to the column and with a small (<< 1 ml) volume loop. Failing that, thread the column up into the transferline and use whatever plumbing you can to give a split at the end of the transfer line. Failing that use a needle through the septum of a conventional inlet, with a narrow bore liner, and run split. Avoid adding extra gas at the GC inlet - its mixing with the sample-laden gas from the headspacer is an additional source of variability. Do not connect the transfer line into the inlet carrier gas feed.

For some reason, headpsacers seem to be the cinderellas of the GC world, despite their being so widely used.

Peter

[phospholipi]

Hello Forum, I found this very amazing forum just by google search and registered immediately.

I'm looking especially at this quote, even we have problems with our Sytem Agilent 6890N+G1888 system with ethanol determination. We have not reproducible area results just for using our SST solution in Headspace analytics RSD is about 6 %. We focused on the vent valve, but here I saw the "aldehyde" writing of moster users don't know, how to check correctly.
I want to know, how to check them correctly.

We had agilent service (a really good guy) in our company to check and he changed the vent valve after 1 whole day of check, even we had problems with not reproduceable results, but we can't say that this was really the problem. Sometimes the G1888 is really good, and after the 20th injection of the same stock solution, he has an anamalous result. We are not able to fix the problem, because some times, the system is really good.

Somebody an idea.

[quote="aldehyde"]
The solenoid valves for vent and vial pressurization get hit with sample during each injection so they are a common failure. This is not difficult to troubleshoot or replace on your own, but most users don't know how to determine if their valves are working correctly and don't understand what happens when they are broken (either no sample, decreased sample, non-reproducible peak sizes). I have seen people continue to use a broken headspace without realizing what is going on (for months), blaming bad chromatography and tiny, non-reproducible peaks on the instrument.

[Peter Apps]

An anomalous result after a series of good ones can be due to the GC cycle time and sample equilibrium time clashing - after a number of runs a sample is supposed to be injecting while the next one is supposed to be loading into the oven or something similar. The headspacer then holds a sample for longer in the oven, or even worse holds a filled loop waiting for a chance tin inject. You have to play with the cycle times and equlilibrium times - make both prime numbers so that they cannot be multiples of one another.

Peter

[phospholipi]

headspacer then holds a sample for longer in the oven, or even worse holds a filled loop waiting for a chance tin inject.

Thanks for replaying, but this is not the fact. The headspace is programmed to inject after a regulated programm. If the GC is not ready, injection occurs anyway. But you see it as a undefined peak with a bad retention time. We had this situation sometimes in summer. But this is not here.

[Peter Apps]

[quote="phospholipi
If the GC is not ready, injection occurs anyway. [/quote]

That is the problem - set the headspacer to wait for the GC to be ready.

Peter

[phospholipi]

That is the problem - set the headspacer to wait for the GC to be ready.

Peter

That's not the problem. The GC is ready!

[Peter Apps]

The problem is that the GC is not ready, not that the headspacer is not ready.

If an injection is made when the GC is not ready the column temperature and/or the gas flows will not be at the values that they should be for an injection. This will produce distorted peaks and shifted retention, which is exactly what you tell us that you see.

Peter

[phospholipi]

Sorry, my fault, I wanted to write the GC is ready.

We have about 0.5 min. GC-ready-time before Headspace injection!

[Peter Apps]

Is it always the 20th (or 21st) sample that gives anomalous results ?

Peter

[phospholipi]

No, there is no regularity




Edit: inserted the word "no"

[Peter Apps]

If the 20th (or 21st) sample always gave an anomalous result, that would be regular would it not ?

Let me rephrase the question; is it the same sample in each batch that gives an anomalous result ? If it is, which sample is it ?

Or to put it another way; which samples give anomalous results ?

What is your sample equilibrium time and your GC cycle time ?

Peter

[phospholipi]

Hello Peter, just corrected the last post!

I want to say, sometimes our instrument is running just fine, some hours later it's worse.
I show you some data (area)

5:30 to 11:33 23:26 to 4:49

MEK Ethanol MEK Ethanol

1. Injektion 28,677 355,342 29,04 349,359
2. Injektion 28,174 339,532 28,496 343,139
3. Injektion 29,557 365,013 28,967 348,839
4. Injektion 28,087 335,18 28,507 342,738
5. Injektion 29,331 366,935 28,444 338,520
6. Injektion 30,064 376,799 28,616 344,934
7. Injektion 28,936 353,785 28,242 335,005
8. Injektion 29,206 361,462 28,123 337,486
9. Injektion 29,425 366,768 28,95 356,111
10. Injektion 30,416 378,274

mean 29,187 359,909 28,598 344,015
RSD % 2,567 3,965 1,140 1,937

First running the second rows, from 23:36 to 04:49 with rsd% 1,9 for ethanol - then after it a new series, the first rows from 05:30 to 11:33 with RSD% 3,965 for ethanol - just doubled. All vials pipetted from the same pool solution (DMSO with MEK and ethanol) from the same operator

Cycle time is 40 min.; equilibrium time 50 min; GC programm run time 35 min. GC ready after 39.5 min.

[Peter Apps]

This repeatability is about the best that you can expect from an off the shelf G1888. If you really have to have better than this the easiest way is to run multiple replicates of each sample and standard, and then work with means.

Peter