Help needed: Methanol tailing EPA8015 GRO by FID separation.

[markf]

Hello,

We're trying to develop a method to quantitate GRO in the range of 10ppm-500ppm in soils by methanol extraction and then headspace injection with FID via a Agilent Select Mineral Oil column.

I've tried the direct headspace method before (Extracting GRO directly from the soil in the headspace vial) and while the method worked flawlessly on standards and even with matrix spikes into clean soil, the method failed to meet recovery percentages in actual field samples, which is very odd indeed.
(Samples were double checked via GCMS method 8260 by calculating total voc values).

Qualititive perliminary tests using methanol extraction did work.
however, when trying to build a calibration curve my low concentrations don't seem to work because of methanol tailing into the retention time window.
(the method works by checking a retention time window, then summing the area in the window)

What can I do?

Currently my starting temperature is 40C for 4 minutes, then ramping up to 120C. My RTW is between 1.2 and 5 minutes, and my methanol tails to about 1.5. all within the 40C isotherm.

My split ratio is set to very low 1:1 in order to achieve the neccessary sensitivity.

I also need to mention that 10ppm 25ppm "look" like they would've worked if I had no methanol tail in the RTW, 50ppm already works fine, but the environmental regulator wants 10ppm to be the LOQ, they might be okay with 25ppm but I don't think they'll settle for 50ppm.

Sorry for the long read, and thanks in advance.

[Peter Apps]

I don't see the point of doing a headspace analysis when the matrix is a volatile solvent - you could just as well (probably better given that the methanol is probably the most volatile thing in the sample) with a straightforward liquid injection with split. You will be able to afford to split because the quantities of analytes will not have been reduced by having them partition into headspace.

A 1:1 split combines all the worse features of split and splitless - the total flow is too low for the EPC to handle and you still get pressure pulses, and the inlet is flushed only bery slowly - leading to the tailing that you see. Better to do a proper 30s to 1 minute splitless injection with a decent total flow (20-30 ml min) to cut the trailing edge off the solvent peak.

Peter

[Peter Apps]

I don't see the point of doing a headspace analysis when the matrix is a volatile solvent - you could just as well (probably better given that the methanol is probably the most volatile thing in the sample) with a straightforward liquid injection with split. You will be able to afford to split because the quantities of analytes will not have been reduced by having them partition into headspace.

A 1:1 split combines all the worse features of split and splitless - the total flow is too low for the EPC to handle and you still get pressure pulses, and the inlet is flushed only very slowly - leading to the tailing that you see. Better to do a proper 30s to 1 minute splitless injection with a decent total flow (20-30 ml min) to cut the trailing edge off the solvent peak.

Peter

[James_Ball]

You are probably limited to what you have available, but the best way to do this would be purge and trap. I can get down to 100ppb GRO in soils with GCMS using purge and trap and FID would be even more sensitive. Using a 5g sample size and concentrating on the trap does wonders for sensitivity compared to headspace alone. The total amount of methanol is what is probably giving you the tailing problems when using headspace.

With your headspace though, you could try injecting at a low split of maybe 5:1, then jumping up to a 100:1 split after enough time has elapsed to transfer the headspace sample to the column. At 6ml/minute at 5:1 split you should be able to transfer 1ml of loop or syringe headspace in less than 30 seconds, then jump up the split after that to help eliminate the methanol tail from any methanol hanging around in the transfer line and injection port.

[markf]

I'm indeed restricted to the equipment I have available, I don't have a P&T apparatus, and I'm using openlab chemstation edition - I have no idea how can I change the split mid-injection.

edit: if it helps, I'm using a PAL CTC autosampler with a 2.5mL headspace syringe.

[Peter Apps]

You can do liquid injections with the CombiPal - just put in the appropriate syringe.

Peter

[markf]

You can do liquid injections with the CombiPal - just put in the appropriate syringe.

Peter

I know how to change syringes, I'm not THAT green I'm refering to the change of split mid-injection - no idea how to do that.

[Peter Apps]

With a liquid injection you will not need a change in split during the run. Trying to do headspace on a solvent extract is just making your life more difficult than it needs to be.

You might be able to contrive a change in split using the gas saver - just set the gas saver to a higher rather than a lower flow.

Peter

[Bigbear]

If you are looking for GRO's and benzene is the first component you are looking for, try upping your starting temp to 60 or so.

Having not done soils ever, why would you need MeOH in the headspace vial?

[markf]

Well, the story is a bit ... embarassing, professionaly speaking, so bear with me, here goes:

We have 2 instruments, one GC/MS and one GC/FID.
The GC/MS is almost always in use in 8260 and 8270 analysis, and the GC/FID is often unused and free to work on.

Thus, when we were asked by the regulator to develop methods to analyse GRO and DRO+ORO, we decided on EPA8015 as the analysis method of choice, and EPA5021 and EPA3550 as the appropriate preparatory methods accordingly.

Both methods were developed rather easily and validated beautifully.

The problem arose when we participated in a proficiency testing round.
The DRO/ORO sample passed with no problems, but the GRO sample passed with a barely satisfactory z-score, having achieved 20% recovery from the assigned value and we were not going to fool ourselves - something in our method doesn't work the way it should.

We were extremely puzzled as to the reasons. Our standards were fine, our work methods were fine, all QC criteria were absolutely okay, including internal QA rounds i.e. blanks, LFB's, matrix spikes etc'.

In the end we smartened up and ran sample duplicates on the MS and on the FID, and there we saw the problem.
The actual extraction from the solid in the headspace vial is not satisfactory, but in the GC/MS, the internal standards added compensate for the low recovery.

Having no internal standards (and no possibility of working with internal standards in this method (remember we're "chained" to 8015 and 5021)),
We were "blinded" to this low recovery from "real" samples, "real" analyte compositions and "real" matrices, until getting the PT results, because the calibration curve was calibrated to these "low" concentrations.

We tested many "allowable" changes to the preparatory method, higher temperature, more agitation, acidic matrix modifier, nothing seemed to help.

In the end we read (in 5021) that for higher concentration ranges, it is allowed to extract the soil samples with methanol and then add a portion of extract direcly to the headspace vial.

As I've said before, perliminary tests and comparisons on medium-high concentrations worked, but now I'm helpless with my low-end of the curve.

*sigh*

And that's why I'm forced to add large amounds (relatively) of methanol to my analysis.
Any ideas on how to fix this? If I knew how I can optimize the headspace preparation I wouldn't have moved on to the high concentration method.

Sorry for the giant wall of text, and thanks in advance.

[Yama001]

I wonder if Carbon Disulfide as an extraction solvent is an option? Not that anyone would want to work with that stuff, but I recall it has a low FID response. Some NIOSH methods used it on charcaol tubes. The methanol tail is tough to get around. A 35 dergee start may help, but I guess the GRO window would move up as well.

[markf]

I wonder if Carbon Disulfide as an extraction solvent is an option? Not that anyone would want to work with that stuff, but I recall it has a low FID response. Some NIOSH methods used it on charcaol tubes. The methanol tail is tough to get around. A 35 dergee start may help, but I guess the GRO window would move up as well.

5021 states methanol. I would've used DMSO if I could since it has a very low response to headspace extraction.

[Peter Apps]

What can you change, and what is cast in stone by the official methods you choose to follow?

There is no point us suggesting things that will work but that you are not allowed to change.

Bear in mind that not everyone knows the details of the official methods.

Peter

[markf]

What can you change, and what is cast in stone by the official methods you choose to follow?

There is no point us suggesting things that will work but that you are not allowed to change.

Bear in mind that not everyone knows the details of the official methods.

Peter

We use EPA methods, for GRO: Analysis method 8015C:
https://www.epa.gov/sites/production/fi ... -8015c.pdf

For prep we use method 5021A:
https://www.epa.gov/sites/production/fi ... /5021a.pdf

Those methods are, of course, free to browse and use.

Page 4 states:
Gasoline range organics may be introduced into the GC/FID by purge-and-trap (Methods 5030 and 5035), automated headspace (Method 5021), vacuum distillation (Method 5032), or other appropriate technique.

*shrug* So, I guess, not direct injection, not that it would likely to help me by much.

Method 5021 clearly states to use Methanol as the basis to all standards and extracts.

[Peter Apps]

"or other appropriate technique."

Means that you can do what you like as long as it works.

Don't take it the wrong way if I tell you that I have better things to do than read all the way through those methods when it is not me who has a problem with them.

Peter