Headspace Peak Shape/RSD% Problems

[modonovan]

Hello,

I am running a headspace residual solvents method on an Agilent 7694A (G1289A) headspace autosampler attached to a 6890A GC. I have connected the column directly to the transfer line ZDV union which gives good peak shape in the standard. However the RSD of the analyte peak areas is around 15% and the RSD of the internal standard ratio is >50%. The internal standard is acetone and I am quantifying chloroform. The sample matrix is purified water for the standards.

If I connect the headspace to the split/splitless gas line and run the inlet in splitless mode (no purge and capped septum purge) with a 1mm ID liner I get very good (approx 3% RSD) peak areas and <1% RSD Int Std ratios. But with this setup I cannot meet some of the method suitability parameters (system sensitivity SNR).

GC Parameters
Column: RTX-WAX (60m x 0.32mm, 0.5um film)
Detector: FID @ 280DegC
Inlet: 110DegC
Oven Temp: 40DegC initial

Does anyone have any idea what is causing the large injection to injection variability in the peak area when the headspace is hooked up directly to the column? Any help would be apprecated as I have been working on this for weeks!

[chromatographer1]

What size is your sample loop?

How much carrier pressure and flow do you use?

What levels of analyte are you trying to measure in your sample and how much sample do you use in your vial?

How much water is in your vial and how big are the vials you are using?

These details can be quite helpful in correcting your procedure.

best wishes,

Rod

[modonovan]

I am using a 250ul sample loop. The carrier is N2 at approx 15psi.

I am using 20ml headspace vials with silicone/ptfe septa filled with 10.0ml of standard.

I don't have the method in front of me, but off the top of my head i think the standard contains 3ug/ml internal std (acetone), 1ug/ml benzene, 3ug/ml chloroform and 1ug/ml 1,2-dce.

I don't think that there is anything wrong with the method itself (which is good as i can't change much), it is more of an issue with the peak area & int std variability when the column is connected directly to the headspace transfer line. I don't understand how plumbing it in a slightly different way can have such an effect.

[Peter Apps]

Depending how you have the pneumatics of the GC and sampler set up, with a direct connection of transfer line to column the carrier gas pressure and flow are controlled by the headspacer. If the transfer line is connected to the GC inlet the 6890 pneumatics also come into play. Setting the split ratio to zero will not necessarily give you a splitless transfer if the pressure in the inlet is higher than is set on the 6890 - the back pressure regulator will bleed gas to atmosphere to keep the inlet pressure at its set value. This would explain why your signal:noise is poor - some of the sample is lost from the inlet.

The poor area repeatability with a direct connection can be due to tailing making integration eratic, or it could be related to vial and carrier gas pressures. 250 ul onto a 0.32 mm id column at nitrogen flow rates gives a wide starting plug which makes the peaks even uglier.

I would go for a split transfer (with the parameters set on the 6890) with a larger loop, and/or a higher sample temperature to get bigger peaks and a passing S:N.

Peter

[modonovan]

I am getting a small reduction in peak area when running it through the inlet, but the main problem is peak efficiency. The peaks are broader when run through the inlet (presumably the extradead volume?) than with the direct connection of the column to the transfer line.

I would like to run it in split mode with a 1ml loop that was already installed in the headspace. However someone in regulatory put the loop size into the submission, so I can change almost every parameter apart from the loop size, oven program, detector settings and standard/sample conc.

The headspace is currently set with the following:

Oven 85DegC
Loop 100DegC
Transfer Line & Inlet 110DegC

25min vial equilibriation in oven
mixing on high

8psi N2 vial pressurisation
0.40min pressurise
0.05min loop fill
0.05min loop equilibriate
1min inject time

From what I have read these temperatures are about ideal for solvents in water (10-20DegC less than bp). The loop fill was modified from the method 0.40min after running a few test runs. The vial equilibriation time has also been optimised.

I don't think it is possible to get more sample on the column by changing the headspace timing/temp so I am mainly looking at better peak efficiency. The direct connection setup meets all of the system suitability parameters apart from injection repeatability (i need 8% RSD for the int std ratio).

[chromatographer1]

I would suggest you use a split with the loop you have and see if works for you. Nothing ventured, nothing gained.

best wishes,

Rod

[modonovan]

I did a quick sequence of splitless & split (ratio of 1:1 and 2:1) as you suggested, got narrower peaks with split presumably due to faster flow through the transfer line but the peak areas were about half those of splitless (on both split ratios) and the snr was slightly below required.

The RSD for the int std ratio for 6 injections were in the range 2-5% for both split and splitless when plumbed through the inlet.

[Peter Apps]

Something is odd - you get good repeatability of peak area ratios, which implies good signal:noise, but it fails the S:N criteria of the method. How are you determining S:N ?, are you sure that there are no peaks eluting during the noise measurement ?

It would be good to see a chromatogram - posting instructions are in a sticky at the top of the LC page.

Peter

[modonovan]

Sorry, wasn't very clear on the last post. Injection of six standards gives a good RSD, injection of a LOQ test solution (1/10 std strength) gives a snr of about 8, but I need 10 to pass system suitability.

The site where the method was developed (on a Varian 3800? with a Varian Genesis HS with transfer line connected directly to column) usually get a snr of 14. I get about the same when the headspace transfer line is connected directly to the column, but then my RSD% for my standards is dire.

I am calculating snr manually using the ep method (range of noise over 20xpeak width at 50%) on a virtually flat bit of baseline. I think it is the signal rather than the noise that is the problem.

I'm not in work again until wednesday afternoon, will try and post a couple of example chromatograms then.

[Tomasz]

What carrier gas pressure do you use for transporting the sample from the loop to the injector?
Have you tried shorter injection time and higher vial pressurisation?

Tomasz

[modonovan]

I use N2 at 15psi (approx flow 2.1ml/min). From what I have learned the injection time is not a factor in the amount of sample that is transferred as the loop is swept in <10sec.
I have not tried varying the pressurisation pressure as it is another thing in the regulatory document, however I have tested reducing the loop fill time and found it to give largest areas at a fill time of 0.05min.

[Peter Apps]

With such a short loop fill you are not bleeding the vial headpsace down to atmospheric when you fill the loop. This makes the pressure in the loop and the peak area vulnerable to small fluctuations in gas flow as the loop fills, hence the poor rsd. The puzzle is why is the rsd better with a connection through the inlet than with a direct connection. How do you have the GC inlet and headspacer gas supplies plumbed ? - on some Agilent headpspacers the sampler gets its gas from the GC or from on-board controllers depending how it is set up. Please post the various pressure and/or flow settings on both the GC and the headspacer for both of your column connection options. Do you see any differerence in retention time between the two options ?

Peter

[modonovan]

I am using the Agilent headspace without the onboard vial pressure regulator or carrier mass flow controller (the G1289A). Vial pressurisation is controlled by an auxillary EPC controller module.

Setup for connection through inlet:

AUX3 Pressure - 8psi (for vial pressurisation)
Inlet flow line from EPC controller cut approx 1inch from the inlet weldment, gas flow routed into headspace, through valve/loop system and then through an 80cm transfer line to a zero dead volume connection just before the gas line enters the inlet. A 1mm ID splitless liner is installed and the column is installed into the base of the inlet as normal. The whole epc-headspace-inlet flowpath was leak checked as per hs manual (capped column nut, capped septum purge). The vial pressurisation pathway test was also completed. Inlet at 15psi const pressure

Setup for direct connection:

AUX3 Pressure - 8psi (for vial pressurisation)
AUX3 Pressure - 15psi (for flow)
Carrier inlet line on headspace plumbed into AUX 5 epc controller. Transfer line connects to column using zdv union with column fitted (column passes out of inlet (110degC) and connects straight to transfer line).

There is slightly longer retention when plumbed in through the inlet (dead vol).

[modonovan]

Found out why it wasn't working - I have been calculating SNR wrong! Is there a good reason why you have to multiply the signal by two in the EP?

[Tomasz]

Actually, you don't multiply signal by 2. You divide noise by 2. The noise is a key here.

Tomasz