Recovery Problems in HS GC

[Mike H.]

I have a 5 component separation consisting of MTBE, MeOH, ACN, IPA and Pyridine. I'm using DB-Wax column 30m x 0.53 id and 1.0µm film thickness. DMSO is the solvent. Sample prep is 100mg sample,2ml DMSO and 1ml sat Na2SO4 in a 10ml HS vial.

I've having problems getting recoveries under 10% for MTBE which is the first eluter and Pyridine which is the last. MeOH, ACN, and IPA are generally less than 10% RDS.

Also when I run the 6 system suitability standards, which contain no sample, RSD for MTBE is <5% and RSDs for the other components including pyridine are ~2.0%.

I was hoping to hear some ideas on how to approach this problem.

[chromatographer1]

Since the problem is worse with the most volatile component when the sample is present, it could be a matter of non-homogeneity of MtBE content in the sample which causes variable amounts of MtBE measured in the preparations containing the sample.

Removing all solvents from a dissolved sample first and then spiking portions of this solution with the solvents of interest should show if this is the problem.

If the low RSD results were performed with the same method that the higher RSD results were achieved, then this indicates that the presence of the sample is the cause.

The factors which must be optimized for the sample preparations are time, temperature, and pressure. These seem to be different than for the blank preparations without the sample.

You may have too short of an equilibration time after pressurization of the vial, or not enough sample is generated to properly flush out the empty sample loop.

Good luck,

Rod

[Johnny Rod]

Forgive the potentially stupid question, btu are you making multilpe injections from the same HS vial?

[Mike H.]

Forgive the potentially stupid question, btu are you making multilpe injections from the same HS vial?

One injection per vial. I read about multiple sample techniques that require log plots, but I'm not there yet.

[Mike H.]

Since the problem is worse with the most volatile component when the sample is present, it could be a matter of non-homogeneity of MtBE content in the sample which causes variable amounts of MtBE measured in the preparations containing the sample.

Removing all solvents from a dissolved sample first and then spiking portions of this solution with the solvents of interest should show if this is the problem.

If the low RSD results were performed with the same method that the higher RSD results were achieved, then this indicates that the presence of the sample is the cause.

The factors which must be optimized for the sample preparations are time, temperature, and pressure. These seem to be different than for the blank preparations without the sample.

You may have too short of an equilibration time after pressurization of the vial, or not enough sample is generated to properly flush out the empty sample loop.
Good luck,

Rod

I'm using an standard Agilent G1888. Do you have any guildlines on what these parameters should be? The method now is:
aux pressure 14 psi
Vial Eq.:15 mins
Vial Pressuration: 0.15
loop fill: 0.50 min
loop eq.: 0.06min
inject:0.50min
oven temp:90°C
loop temp 180°c
transfer temp 180°C

One more question that come to mind. I'm using a salting out technique for the prep, but the sample is difficult to get into solution and may not always be in solution prior to the salt addition. Might this cause a problem?

[chromatographer1]

As the most volatile component has the greatest RSD value, this indicates an equilibration or leak issue. I wonder what pressure the DMSO/water solution has in the vial after heating at 90C for 15 min and if the vial is sealing consistently?

Your valve/loop temperature is high (180C). Is your rotor failing to seal? Is it scratched or could the valve be leaking at a joint of the sample lines?

I would not expect the MtBE to have a higher RSD than pyridine, unless you are splitting the sample in the injector and are getting discrimination there.

2% RSD values are good acceptable results (not great, but with 3mL of liquid in your vial with only 15 min equilibration time it is not bad), but MtBE should give you better not worse results unless you are saturating the HS volume with MtBE which seems unlikely.

This leads me to think that leaking hardware (including the vial itself) might be the primary cause. Perhaps you sealed your std vials well and then sealed your sample vials less well. Does this performance duplicate, triplicate?

Good luck in finding the cause of your poor results.

Rod

[Mike H.]

If I had a leak somewhere wouldn't it affect my sys suit injections? I test 6 sys suit preps consisting of 2mls wrk std and 1ml Na2SO4. These do not contain any sample and I get acceptable RSDs for all components. <5% for MTBE and <2% for the rest.

Generally what do you pressurize your HS vials at? How long is loop pressization and loop fill?

This situation is weird because I get acceptable sample solvent recoveries from MeOH,IPA and ACN <10%RSD (peaks 2-4) but fail recovery ~20%RSD for MTBE and pyridine (peaks 1,5).

[chromatographer1]

I pressurized the vials at a pressure greater than the pressure generated during the heating equilibration process, and with enough gas in the vial to push all the carrier gas out of the transfer lines through the loop and then purge the loop three times at a minimum.

You have to determine this yourself as I have no idea of the dead volume in your system.

Generally when I used by HS analyzer I pressurized between 5 and 15 psi. If your vial pressure is low when you add an additional 15 psi then it might take a while to reach a re-equilibration. The most volatile component should be the one with the greatest problem with equilibration, and your results are........ yes, the most volatile has the greatest variation.

Apparently, the presence of the drug makes a big difference with the equilibration. Without knowing the actual areas of the MtBe peaks of the sample preparations compared to the std preparations trouble shooting is difficult. Does the presence of the sample increase or decrease the response? What do these indications suggest to you what might be the nature of the problem?

You see there is a lot of information missing which you have not given (probably for good reasons) which makes this a guessing game.

Vary your conditions until you get consistent results. Research is always a bother but sometimes we have to do it to find causes for problems.

best wishes,

Rod

[Bigbear]

We had the same sort of issue with a new HS system doing Blood ETOH. Tried to use the same parameters we used for the old system. It was a temp. problem in the headspace unit ( we were running too hot).
Ended up incubating @ 65, valve oven @ 70 and transfer line @ 75.

[Mike H.]

I pressurized the vials at a pressure greater than the pressure generated during the heating equilibration process, and with enough gas in the vial to push all the carrier gas out of the transfer lines through the loop and then purge the loop three times at a minimum.

How do you know when you've purged the loop exactly 3 times? Is there an equation to determine this listed somewhere?

You have to determine this yourself as I have no idea of the dead volume in your system.

I'm using a G1888 headspace from Agilent. An Agilent technican setup up our system so everything should be default. Can void in the HS be estimated assuming this condition? If not how can the void volume in the headspace unit be determined?

[chromatographer1]

You learn the length and ID of the tubing, including the sample loop and determine the volume of this path.

Then you create enough sample to exit your vial to flow to the sample loop, plus 3 times its volume. So if the volume of tubing from the vial to the loop was 1 mL , and the loop held 1mL then you would need 4mL of sample to leave the vial in order to purge the sample loop three times.

best wishes,

Rod

[Mike H.]

The headspace parameter are all in time units of minutes. So if my loop fill was 0.2 minutes, how is this converted to volume?

[chromatographer1]

There is no conversion of fill time to sample loop volume. It is like asking if I am going 60 km/hr in a automobile how wide is the car?

If you have a timed injection then there is no sample loop. (PE and others)

If you have a timed injection with a valve and loop then that time is when the loop is purged onto the column and has no basis in the calculations for the dead volume of your sample system. (Agilent and others)

If you have a valve operated sample loop then determine the loop length and ID and determine its volume. There may be a tag on the loop giving your its internal volume.

best wishes,

Rod

[Mike H.]

I'm using an Agilent G1888 coupled with an 6890 GC so void volume isn't a concern. Thanks you've been a big help.