Chromatography Forum

Hi everybody,

I'm currently working on trace analysis of benzene in vegetable oils and animal-fat matrixes by GC-MS-Headspace. We have to reach an LOQ of 0.01 mg/kg.

In a NaCl-saturated aqueous solution the method works fine, but (as expected!) when we passe to a fat matrix (vegetable oil or animal fat) we obtain very bad results.

I already found an article where authors worked on vegetable oils for benzene analysis and they announced LOQ near our goal (so in theory it would be possible to improve our method!), but they didn't specify detailed incubation information for HS vials. I already sent a email bu no answer until now

Our conditions are:

5g of sample in a 22-ml HS vial are incubated in a GC-PAL system for 30 min at 75°C with shaking speed set at 300 rpm; after this, 1ml of headspace is injected in GC-MS with a DB-624 type column. The internal standard used is benzene-d5.

MS detector is set to SIM mode.

Do you have any ideas which coul help me to improve performances of this method, please?

Or maybe such a LOQ is unrealistic for this kind of matrix?

I thank you in advance for your valuable help!

Best regards,

Carmen

Did you try their conditions?

I just tried to measure benzene in sunflower oil on my static headspace system. I only have a flame detector on that instrument but given what I expect the signal-to-noise to be for a minimum quantitation amount, I think I could see 35 ppb or so reliably. This particular sample of sunflower oil contains less benzene than that. With an MS detector in SIM mode, I could likely go lower.

2.00 g sample. Incubate at 60 °C for 20 minutes. Inject. 75 °C should work - perhaps better.

I used the method of standard addition to calibrate. I created my stock standard in the sample (neat sunflower oil). I made 35 µg/g benzene in the sunflower oil as my stock standard. I added 0.2 µL of benzene (density 0.874 g/mL, Sigma-Aldrich) to about 2 g of sunflower oil then diluted it to 5.00 g with more sample. If I wanted to be super accurate, I'd measure out a lot more benzene into perhaps a lot more sunflower oil. You can do this however you want to minimize your errors. I just wanted to prove concept.

After that, I used the unfortified sample to dilute the standard to 2.00 g in my vials. Add 10 and 30 mg of standard and dilute both to 2.00 g with more sample.

Making your standard in the sample pretty much eliminates extra components that are not in the matrix. You could try the same technique on some other oils to verify that the calibration data is what it is in pretty much all vegetable oils.

A quick analysis on a different instrument with a MS detector (SPME sampling) shows that I have about 12 ppb (ng/g) benzene in this sunflower oil.