Chromatography Forum

I am having a problem choosing appropriate parameters for a residual solvents analysis using the headspace autosampler and a splitless GC method.

I have no choice about the basic parameters. We would not ordinarily choose to use a splitless headspace method for a residual solvents method we developed ourselves, but this method is being transferred from a different lab, and unfortunately has to be made to work substantially intact.

Here are the parameters:

Instruments:

Agilent G1888 Network Headspace Sampler
HP 6890 GC System

Column:

Phenomenex ZB-624, 30 m x 0.53 mm ID x 3 um film thickness; max T = 260 C (Isothemal/Program)

Solvents to be tested (level in standard in brackets):

MeOH (600 ppm), EtOH (1000 ppm), IPA (1000 ppm), CH2Cl2 (100 ppm), Methyl t-butyl ether (MTBE) (1000 ppm), DMF (200 ppm), Toluene (200 ppm), 1,4-Dioxane (35 ppm), 2-Methoxyethanol (a.k.a ethylene glycol monomethyl ether, EGME) (5 ppm).

This is made up in DMSO.

GC method:

Inlet: 170 C, 2.2 psi, Purge Flow 22.0 mL/min, Purge Time 0.60 min. Gas saver not on. Total Flow Setpoint = 27.2 mL/min
Oven: 40 C hold for 9 min. Increase to 200 C at a rate of 20 C/min, hold 6 min. Total = 23.00 min
Detector (FID): 220 C, H2 30 mL/min, Air 400 mL/min, Makeup (constant, N2) 40 mL/min, Lit Offset 1.0

Headspace method:

Oven: 140 C
Loop: 150 C
Transfer Line: 155 C
Vial Equilibration: 30 min
Pressurization: 0.50 min
Loop Fill: 0.50 min
Loop Equilibration: 0.50 min
Inject: 0.12 min
GC Cycle: 35 min
Carrier Pressure: 4.5 psi
Vial Pressure: 10 psi

Things I Don't Have a Choice About:

The following I would prefer to change only as a last resort because I need the method to be transferred essentially as-is:

The method has to be headspace, and unfortunately must be splitless. The inlet pressure and temperature are specified in the transferred method, as is the oven program and some of the headspace parameters. The temperatures chosen are those from the transferred method, as are the times. The column comes from the transferred method.

Things I Have A Choice About:

The lab that developed this method used different instrumentation, which I don't have immediately available to tell you but I can check if necessary. They didn't specify anything for parameters such as Purge Flow and Purge Time, and we had to make educated guesses about how some of their headspace parameters mapped to those used by the Agilent G1888, especially the times.

Observations:

It is necessary to adjust the Carrier Flow knob on the HS. Too high and the inlet pressure shoots up and cannot be maintained at 2.2 psi. I have tried to set the carrier flow to the maximum that will allows the 2.2 psi inlet pressure to be maintained.

The Problem:

I have managed to get the method to the point where I can see all of the peaks except for the 2-methoxyethanol peak in the standard. The transferring lab only gets a very tiny peak for this but it is below the limit of detection with the last method I used (I had 50 mL/min purge flow and 0.06 min purge time for that).

From that point I was trying to make adjustments, primarily to the purge flow and purge time, because these are parameters that weren't specified so I have the greatest leeway with, to get the sensitivity to the point where I can see the 2-methoxyethanol.

When I tried to run it this time though, I am getting no peaks at all. When I manually inject 1 uL of methanol I see a huge peak come out on the online plot, so I don't suspect anything of being wrong on my GC side. I found a post saying that the flow ought to be between 20 and 30 mL/min for a method similar to this, so I tried lowering my purge flow and increasing my purge time to the paramters above and still no peaks (except a tiny DMSO peak).

We don't have a lot of experience with splitless methods in our company, especially using headspace, and I wonder if anyone can offer any insights as to what I'm doing wrong.

Thanks!

I should also have said, mode is constant pressure. Carrier gas is He.

The purge time and flow are parameters of cleaning the sample flow path and sample loop, correct?

That should not effect the sensitivity of the analysis unless there is enough active sites in the tubing that absorb your solvent peaks.

I would ask questions concerning the original developer of the method:

What kind of tubing did they use in their sample path. Did they use a fixed loop injection method?

I would also confirm the flow rate through the column by injecting a vial of methane or natural gas through the HS analyzer. Don't believe digital readouts to absolutely true. They can be adjusted, and can be adjusted incorrectly.

What model of HS analyzer was this method developed. What happens if you are not able to transfer the method using the Agilent hardware?

good luck and best wishes,

Rod

The originating lab instruments were as follows:

Gaschromatograph PE Autosystem XL with Headspace HS40 XL or
Gaschromatograph Clarus 500 with Headspace TurboMatrix HS40

The method parameters we were given were as follows:

GC:

Oven temperature program: 40 °C, 9 min, 20 °C/min, 200 °C, 6 min
Inlet temperature: 170 °C
Detector temperature: 220 °C
Carrier gas and flow rate: Helium, 2.2 psi
Air flow: 400 mL/min
Hydrogen flow: 30 mL/min
Split ratio: splitless
Run time: 23 min
Aux 3: 25.0 psig (3 valve ON)

HS:

Oven Temperature: 140 °C
Needle Temperature: 150 °C
Transfer Temperature: 155 °C
GC Cycle-Time: 35 min
Thermostating Time: 30 min
Pressurization Time: 0.5 min
Inject Time: 0.12 min
Withdrawl Time: 0.5 min
Vial venting: OFF


I'm not sure what kind of tubing they used but I can check...

We don't have much (almost no) experience with splitless methods... I am not even properly sure what these parameters really do, although the ChemStation help files do provide some information and suggestions that I have tried to exploit.

If I cannot transfer the method as-is in a fashion that it allows me to at least detect the 2-methoxyethanol, the next avenue is probably to develop a split method, which is something I have much more confidence with, but the slight complication is that this is a validated method and I'd like to just perform a method transfer validation as opposed to validating a new method from scratch. Time is an issue .

I am sorry to tell you that you are not likely to get the same results with the Agilent HS analyzer. It uses a fixed loop design. The patented Perkin Elmer design uses a timed injection method and it allows the direct sampling of the vial into the column which is within the transfer line with almost no dead space or active sites to deal with.

The PE injects a volume equal to the flow rate times the time injected of a pressurized gas sample. You may not be injecting the right amount of sample using the fixed loop design.

In any case it is possible that you cannot transfer the method as the sampling hardware designs are too different. Best case will be that it will be difficult.

You will need a deactivated loop and transfer lines, a direct connection to your column, and possibly some 'wash' samples to clean off your rotor carryover for some analytes, especially ester and nitrogen containing analytes.

You do have a task ahead of you and I hope you do not find it too difficult.

best wishes,

Rod

Well, I thank you for your insights, this information, while not exactly encouraging, is very useful to us. We will try some further experiments and decide what we want to do.

Any further feedback from anyone on how to choose the purge flow and purge time parameters is also greatly appreciated - even if it does not turn out to have a direct bearing on our problem it would at least be educational.

Adjust your purge flow and time so that no residue of DMSO or any of the test analytes are present when an empty vial is processed through your analyter.

If my understanding is correct, this purge flushes out the needle and transfer line of the sample as it passes through the sample loop to the vent

When I developed HS methods using a similar Tekmar 7000 I used direct injection and my carrier flow was anywhere from 5cc to 20cc/min.

My purge flow was 20 cc/min. and was continuous.

I kept my sample valve in the inject position to flush the loop with carrier gas for at least 30 sec. Longer does not hurt. You can reset the valve to fill the loop right before actually sampling the next vial.

best wishes,

Rod

Thanks for the advice - I still haven't found parameters to completely get rid of the carryover when I inject a blank vial (availablility of the instrument has become scarce recently) but I did manage to see all of the peaks I need to which is great.

My fear though is that the method is still not really transferrable; my feeling is that if we were to attempt intermediate precision that it would not pass, and that there is an element of luck as to whether the method is sensitive enough from one run to the next. I hope I'm wrong about that, but if it turns out I'm right, I'll have to develop a split method (where I'm on firmer ground) and do a full validation. Oh well...

I can understand you would like to be on familiar ground when developing a method. Certainly it is not uncommon to use a fixed loop injector and a split injection when using capillary columns. But I think that you should not blame the splitless methodology as that isn't the problem.

When I was developing methods on HS40 and the Tekmar 7000 I was able to reach the lowest detection limit with the fixed loop Tekmar, but the difference was not great. The timed injection method of Perkin Elmer had its advantages of minimal carryover and active sites but was limited by the peak width of the injection for high resolution analyses. The Tekmar was limited by the same limitation due to large loop volumes needed to reach very low limits.

Both limitations were overcome to a large degree by increasing the headpressure of the inlet where the sample was taken. This is achieved by increasing the restriction of the column either by using a smaller bore and longer column or by butt connecting a short piece of 0.25mm capillary behind the analytical column just in front of the detector. I would use a 1 to 5 meter piece of inert FSOT, usually uncoated but deactivated.

If your column carrier flow is set too low you can have problems with splitless injection. You do not need to be at the optimum carrier flow rate for most capillary headspace analyses. I used to perform the five component OVI USP test in 2.5 minutes isothermally. Talk about throughput ! It took longer to heat the samples than to do the chromatography. It was silly to do a presence test with a resolution of 6 to 12 when a resolution of 2 was adequate for measurement.

I have mentioned my HS article published in Analytical Chemistry back in June 1997 several times in the Forum. You will find that your detection limit of difficult analytes like dioxane and EGmME (2-methoxyethanol) will improve if you use deactivated sample transfer tubing, a large 0.5-1.0mL sample loop, a faster column flow rate (5 to 20cc/min) COMBINED with a small liquid sample in a smaller headspace vial. With these conditions I was able to see 1ng of residual solvent in a 3.35mL vial using 25µL of sample.

In any case I hope you are able to measure your analytes at the desired levels using whatever method you chose. I suspect you will have to perform a new validation for your hardware with either course you choose, fixed loop with split or splitless injection.

best wishes,

Rod

Hi Rod,

Thanks for your interesting and informative response. I haven't been able to get back to this particular problem yet in the last week or so, but I will have to tackle it again soon, and the more information I have to go on the better.

Stephen

maybe this can help

http://www.restek.com/aoi_pharm_A017.asp