Agilent G1888 Headspace.

[mburleson]

We were attempting to replicate a paper analyzing hand sanitizer for residual solvents/adulterants and observed something I believe that could be an issue of the headspace autosampler. I was having my students use cyclohexane for recovery calculations. They spiked their samples so that the concentration would double, but it did not. Some groups saw about a 1.2 increase in response and saw a decrease in response.

One thing that I did notice, is that the paper didn't specify a vial pressurization or carrier gas pressure. Right now our system is "set" to 15 and 25 psi, respectively. I put set in quote marks because that's what the headspace method is. However, when I checked the actual pressures on the headspace, I observed them to be 14.8 and 17.3 (vial pressurization and carrier).

Our GC uses EPC to control vial pressurization and MPC for carrier pressure so that we can disconnect the transfer line from the back inlet. The GC uses an EPC to supply the carrier pressure and is set to 65 in the method/instrument. The He cylinder is set to an output of ~80 psi.

Ignoring that they're students and prepared everything correctly (ha), this has me concerned. I've come up with the following plan:

Do manual injections to see if the area doubles.

Check for leaks on the HS.

Repeat the analysis using a different spike.

Use methanol, acetone and 1-propanol to assess injector linearity.



Is there anything else I can do to assess the headspace?

[rb6banjo]

It sounds like you are running at constant pressure for your separation. The head pressure there is dependent upon the flow you want through the column. I have a 60 m x 0.32 mm x 0.5 µm column and the carrier gas is hydrogen so I have about 12.0 psig as the head pressure. This is 46.5 cm/s linear velocity at 40 °C. I run at constant pressure. If you use the Restek flow calculator (online for your use), you'll see that puts me in the sweet spot for H2 as the carrier gas and this column at 40 °C.

For our headspace sampler, as far as vial pressurization goes, we're supposed to be 3-5 psig for the larger than the carrier gas pressure so I have that set at 16.0 psig.

These settings give me very good precision in my measurements. My samples are substantially water but can contain from 5-15% ethanol.

Hand sanitizer is pretty high ethanol concentration. I don't know what your impurity concentrations are but with such a volatile matrix, you may want to try a very small sample size and the full evaporation technique to eliminate the matrix. That or a low incubation temperature for your sample. If it's too warm, you might be over pressurizing the sample in the vial. If it's not sealed well, your precision will be all over the place because your sample is leaking out.

Full evaporation isn't very good for ppb-type concentrations because your sample size is so small (10-100 µL). But if you're looking for ppm-type concentrations, you might be successful.

[Peter Apps]

Your first step needs to be to estabish whether you get repeatable peak areas when you inject from multiple vials of the same (spiked sample) - if the areas are not repeatable then you will not get linearity vs concentration.

Peter

[mburleson]

It sounds like you are running at constant pressure for your separation. The head pressure there is dependent upon the flow you want through the column. I have a 60 m x 0.32 mm x 0.5 µm column and the carrier gas is hydrogen so I have about 12.0 psig as the head pressure. This is 46.5 cm/s linear velocity at 40 °C. I run at constant pressure. If you use the Restek flow calculator (online for your use), you'll see that puts me in the sweet spot for H2 as the carrier gas and this column at 40 °C.

For our headspace sampler, as far as vial pressurization goes, we're supposed to be 3-5 psig for the larger than the carrier gas pressure so I have that set at 16.0 psig.

These settings give me very good precision in my measurements. My samples are substantially water but can contain from 5-15% ethanol.

Hand sanitizer is pretty high ethanol concentration. I don't know what your impurity concentrations are but with such a volatile matrix, you may want to try a very small sample size and the full evaporation technique to eliminate the matrix. That or a low incubation temperature for your sample. If it's too warm, you might be over pressurizing the sample in the vial. If it's not sealed well, your precision will be all over the place because your sample is leaking out.

Full evaporation isn't very good for ppb-type concentrations because your sample size is so small (10-100 µL). But if you're looking for ppm-type concentrations, you might be successful.

Thanks for your insight!

I think the method is set to be constant flow, but will double check. I used the flow calculator by plugging in our 1.5 mL/min flow rate and obtained the results below.



Our vial pressure is also set to 16 psi for the headspace and the carrier is 25 psi. Our Agilent engineer set it to those and was onsite yesterday to verify. His injection %RSD was like 0.2, but he used the little capillary samples Agilent does. I used DMSO as the solvent and tried using cyclohexane and obtained the results below.



Based on the paper we're attempting to follow (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8762429/), the limits for the impurities are quite high - 0.06 to 0.44%. We're testing the hand sanitizer for acetone, methanol and 1-propanol.

We used the parameters as close as we could from the paper, but we have a G1888 HS, so its parameters are shown below (there's a typo, the split ratio matched that of the paper, 100:1).



I can't figure out what is going. I have done HS when I was industry, and we did system suitability of course, but I don't remember struggling this bad. We always used DMSO as our solvent for the API.

Any advice?

[Peter Apps]

At a temperature of 125C the pressure in the vial is probably higher than the 15psi pressurization setting, and so as the headspacwer tries to pressurize the vial you will get vial headspace backflushing into the plumbing. Reduce the vial temperature to less than 100C.

[mburleson]

At a temperature of 125C the pressure in the vial is probably higher than the 15psi pressurization setting, and so as the headspacwer tries to pressurize the vial you will get vial headspace backflushing into the plumbing. Reduce the vial temperature to less than 100C.

Makes sense! Will do this today and reply. Thank you!

[rb6banjo]

I'll state up front that I would likely do this analysis using a direct injection technique. You have a volatile matrix and volatile analytes. Static headspace analysis doesn't seem to be warranted here.

Like I mentioned above and Peter reiterated, I think 125 °C is excessive for your sample incubation temperature - especially if you have a lot of sample in the vial. The boiling point of your matrix is essentially 78 °C - maybe a little higher because of the ~15-20% water. To do good headspace analysis, all you need is a temperature that provides a reproducible vapor pressure for your analytes in the vial containing the sample AND gets enough analyte into the vapor space that you can get the sensitivity you need.

I took the liberty of quickly analyzing some hand sanitizer in my building for acetone on my headspace system and I chose the parameters I'd choose if the problem were presented to me and I had to use static headspace sampling. My system needed a workout anyway. I first tried methanol but with the wax column I have in my instrument, I can't separate ethylacetate from methanol AND my hand sanitizer has a fair amount of ethylacetate in it as well. Acetone was easier.

0.20 g of sample in the 22 mL headspace vial.
Incubate the sample at 50 °C for 20 min.
Inject

My system is a Perkin Elmer system so the other parameters won't mean much to you. It is also a flame detector. The transfer line from the HS unit goes directly into a split-splitless inlet and there, the sample is split 8:1. Hydrogen carrier gas. My GC column is a Heavywax (Agilent), 60 m x 0.32 mm x 0.25 µm

Take the density of acetone as 0.791 g/mL (Sigma-Aldrich).

Analyze the sample in triplicate (unfortified).
Two other vials prepared by adding approx. 0.1 g of sample then 5 µL and 10 µL of neat acetone to the separate vials, then adding more sample to get up to 0.20 g. Mix well and analyze. Could do a better job by actually weighing the acetone into a known mass of sample and then using that as the stock standard to dilute with more sample. I just wanted to be fast here to demonstrate feasibility.

0.005 mL x 0.791 g/mL/0.20 g x 100 = 1.98% (by weight)
0.010 mL x 0.791 g/mL/0.20 g x 100 = 3.96% (by weight)

Analyzed all of them using this headspace method.

https://i.postimg.cc/Vk4dHrZT/Sanitizer-Summary001.jpg

The image shows what my chromatography looks like, as well as the method-of-standard-addition data manipulation.

Too much sample in the vial + too hot = potentially leaking vials and as Peter mentioned, improper pressurization of the vials. All results in precision that is less than desirable.

In the end, I determine that my sanitizer contains 16.8 +/- 0.3 ppm (standard deviation of the mean) acetone. This sample is well within spec for acetone. Knowing what I know now, I'd go back and fortify the samples at lower concentrations. But, this wasn't too bad and pretty good sensitivity at lower temperature.

[mburleson]

I'll state up front that I would likely do this analysis using a direct injection technique. You have a volatile matrix and volatile analytes. Static headspace analysis doesn't seem to be warranted here.

Like I mentioned above and Peter reiterated, I think 125 °C is excessive for your sample incubation temperature - especially if you have a lot of sample in the vial. The boiling point of your matrix is essentially 78 °C - maybe a little higher because of the ~15-20% water. To do good headspace analysis, all you need is a temperature that provides a reproducible vapor pressure for your analytes in the vial containing the sample AND gets enough analyte into the vapor space that you can get the sensitivity you need.

I took the liberty of quickly analyzing some hand sanitizer in my building for acetone on my headspace system and I chose the parameters I'd choose if the problem were presented to me and I had to use static headspace sampling. My system needed a workout anyway. I first tried methanol but with the wax column I have in my instrument, I can't separate ethylacetate from methanol AND my hand sanitizer has a fair amount of ethylacetate in it as well. Acetone was easier.

0.20 g of sample in the 22 mL headspace vial.
Incubate the sample at 50 °C for 20 min.
Inject

My system is a Perkin Elmer system so the other parameters won't mean much to you. It is also a flame detector. The transfer line from the HS unit goes directly into a split-splitless inlet and there, the sample is split 8:1. Hydrogen carrier gas. My GC column is a Heavywax (Agilent), 60 m x 0.32 mm x 0.25 µm

Take the density of acetone as 0.791 g/mL (Sigma-Aldrich).

Analyze the sample in triplicate (unfortified).
Two other vials prepared by adding approx. 0.1 g of sample then 5 µL and 10 µL of neat acetone to the separate vials, then adding more sample to get up to 0.20 g. Mix well and analyze. Could do a better job by actually weighing the acetone into a known mass of sample and then using that as the stock standard to dilute with more sample. I just wanted to be fast here to demonstrate feasibility.

0.005 mL x 0.791 g/mL/0.20 g x 100 = 1.98% (by weight)
0.010 mL x 0.791 g/mL/0.20 g x 100 = 3.96% (by weight)

Analyzed all of them using this headspace method.

https://i.postimg.cc/Vk4dHrZT/Sanitizer-Summary001.jpg

The image shows what my chromatography looks like, as well as the method-of-standard-addition data manipulation.

Too much sample in the vial + too hot = potentially leaking vials and as Peter mentioned, improper pressurization of the vials. All results in precision that is less than desirable.

In the end, I determine that my sanitizer contains 16.8 +/- 0.3 ppm (standard deviation of the mean) acetone. This sample is well within spec for acetone. Knowing what I know now, I'd go back and fortify the samples at lower concentrations. But, this wasn't too bad and pretty good sensitivity at lower temperature.

Thank you! This is why I used to try things on my own before just replicating everything from a paper.

[rb6banjo]

I feel your pain. I've been down that road a few times too in 35+ years of doing this stuff. You try to replicate the procedure in the paper and it just doesn't work. Lots of times, however, with some modifications, you can get the concept to work.

Best of luck with your little sanitizer project.

[mburleson]

I feel your pain. I've been down that road a few times too in 35+ years of doing this stuff. You try to replicate the procedure in the paper and it just doesn't work. Lots of times, however, with some modifications, you can get the concept to work.

Best of luck with your little sanitizer project.

Thank you! One last question, do you ever encounter higher baselines using H2 as the carrier gas and doing FID? We have a hydrogen generator for our front inlet to our MSD, but our back inlet for the FID still utilizes He.

[rb6banjo]

No. The flame uses 30 cc/min of the same H2 as a support gas. The makeup gas to the flame is N2. The signal is extremely quiet. H2 is a lot cheaper than He. You can see from my chromatogram that S/N on 17 ppm acetone is very good. There are also some other even more trace components in there that show good S/N.

I've been reluctant to change my systems that have an MS to H2 because from everything I've read/heard, you lose sensitivity when you do that. To retrofit the detector for H2 also costs money. Most of the time, I'm looking for microsniffs of stuff so I can't sacrifice any sensitivity.

To me, it's a no-brainer to change to H2 carrier if your detector is a flame (FID or FPD, probably an ECD too). The MS detector is the one that you have to think about.

You'll want to make sure that your instrument vendor says it's ok to simply swap the gases. They may have reasons why you shouldn't or they recommend that you do some other type of retrofit to make it work safely.

[environeer]

Thank you! One last question, do you ever encounter higher baselines using H2 as the carrier gas and doing FID? We have a hydrogen generator for our front inlet to our MSD, but our back inlet for the FID still utilizes He.

As a matter of fact, yes.
For our FID Instrument (Agilent GC 7890A) we switched from bottled hydrogen to a set of generators (PEAK), including hydrogen, synth. air, and nitrogen.
The baseline went from up ~ 4 pA (bottle) to around ~19 pA (generators). This was a very stable increase and therefore “just” an offset, we lived with it.

[mburleson]

Thank you! One last question, do you ever encounter higher baselines using H2 as the carrier gas and doing FID? We have a hydrogen generator for our front inlet to our MSD, but our back inlet for the FID still utilizes He.

As a matter of fact, yes.
For our FID Instrument (Agilent GC 7890A) we switched from bottled hydrogen to a set of generators (PEAK), including hydrogen, synth. air, and nitrogen.
The baseline went from up ~ 4 pA (bottle) to around ~19 pA (generators). This was a very stable increase and therefore “just” an offset, we lived with it.

Thank you for sharing your experience with the swap! I know from our Agilent engineer that anything below 25 pA would pass their specs.

[mburleson]

No. The flame uses 30 cc/min of the same H2 as a support gas. The makeup gas to the flame is N2. The signal is extremely quiet. H2 is a lot cheaper than He. You can see from my chromatogram that S/N on 17 ppm acetone is very good. There are also some other even more trace components in there that show good S/N.

I've been reluctant to change my systems that have an MS to H2 because from everything I've read/heard, you lose sensitivity when you do that. To retrofit the detector for H2 also costs money. Most of the time, I'm looking for microsniffs of stuff so I can't sacrifice any sensitivity.

To me, it's a no-brainer to change to H2 carrier if your detector is a flame (FID or FPD, probably an ECD too). The MS detector is the one that you have to think about.

You'll want to make sure that your instrument vendor says it's ok to simply swap the gases. They may have reasons why you shouldn't or they recommend that you do some other type of retrofit to make it work safely.

We swapped to H2 as the carrier for the back inlet with the FID and HS after getting the okay from Agilent. Thank you for that! It came in handy because our gas supplier sent us a "full" tank of He that was actually empty.

One thing I noticed though, is that I occasionally hear hissing from the back inlet split vent. Because our hydrogen generator can't keep up, the HS carrier gas is supplied directly from a UCG cylinder. Our flow rate is 1.5 mL/min to give us the velocity of 40 at 40C with a column head pressure of ~8. Our carrier pressure for the HS is set to double, so 16. Is that too much? I always went by 3-5 psi higher than the column head pressure but I've been having debates on that.

[rb6banjo]

Sorry. I don't know what to say about this. My carrier gas pressure is supplied through the Perkin Elmer headspace sampler to my split-splitless inlet that feeds the column. My split ratio of 8:1 is governed by the EPC for that inlet.

My system has a 60 m x 0.32 mm column that requires 12.0 psig to get the flow I want (running at constant pressure). My headspace sampler is set to 16.0 psig as what they call "carrier pressure". My understanding is that the 16 psig is required to pressurize the vial and ensure reliable transfer of the equilibrated sample vapor from the vial to the column. Additionally, I need 35 cc/min H2 (350 cc/min air) to support the flame. N2 is my makeup gas.

I do not use a generator at this point. All of my H2 carrier gas comes from a bottle. The secondary gauge on the regulator is set to 80 psig. Plenty of supply for all of the demand.