Ethanol analysis with poor reproducibility in headspace

[peedrohra]

Hello, I am analyzing ethanol samples in aqueous matrix. However, I am not having good reproducibility. RSD% is close to 20% for ethanol. I am using dichloromethane as internal standard and this compound has 2% RSD. I believe I performed over 100 injections. Replicates are never reproductive.
I have already performed factor analysis of various parameters, but found nothing that is statistically significant. I tested the same method on other equipment, the result was not satisfactory. The peaks are well resolved, the problem is that the variation in ethanol area is so "random".
I don't know what it can be anymore. I don't know much about GC, I'm still learning a lot. Therefore, I count on your understanding and help.
---------------------------------------------------------------------------------------
Follow the method used:
Method: Headspace
Injector Temperature: 120°C
Syringe Temperature: 100°C
FID: 120ºC, air/hydrogen: 400:40
Incubate time: 20 min
Incubation temperature: 90°C
Column: SH-RTX-624 - low/mid polarity - 30,0m;0,32mm;1,8um
Sample volume: 500 uL
Split: 5:1
Linear velocity: 35 mL/min
Carrier gas: He

---------------------------------------------------------------------------------------
Sample prepare:

Stock solution ethanol: 2,5 mL ethanol P.A to volumetric baloon 25mL with ultrapure water;

Stock solution Dichloromethane: 150 uL to volumetric baloon 50mL with ultrapure water;

Intermediate Stock Solution Dichloromethane: 2mL stock solution
dichloromethane to volumetric baloon 50mL with ultrapure water.

Replicates 100ppm v/v (5): 50 uL solution stock ethanol and 2mL intermediate stock solution dichloromethane to volumetric baloon 50mL

----------------------------------------------------------------------------------------
We have already changed several of these parameters, just do not change the column and matrix.
I have the impression that it is something related to the organic composition of ethanol. Maybe adsorption interaction with some component, though, don't understand much about it yet. We didn't change the internal default either. Maybe if we change the internal standard for an alcohol, maybe the sample and the internal standard can suffer the same interference ...

[Bigbear]

I feel your incubation temp is too high, I think I used 70C, plus your FID should be at 300 or 350. At 100 you may be condensing water in the detector.\
It's been so long that I can't remember what we used for internl and surrogates, but they were alcohols.

[Bigbear]

I found some old SOP's and these are the settings we used
Head space
incubate at 65C for 24 min
GC
injector @ 125
Detector @ 250
Initial temp 35C initial time 1.5 min, 5C/ min to 65, 60C/ min to 175

Water is not your friend, the incubation @ 65 is enough for the alcohols with out a ton of water . The GC ramp to 175 is to get water out of the column and detector.

[James_Ball]

The solubility of the dichloromethane in water is very low, while for ethanol it is very high. Dichloromethane should always recover very well while ethanol may not, so it is probably not the best choice for internal standard. Maybe isopropyl alcohol would be a better choice for internal standard.

[peedrohra]

BigBear and James_Ball, thanks for the support. I believe the issue has been resolved. I changed the internal standard, I strongly believe that Ethanol and other alcohols suffer some adsorption interaction, either with the syringe or with the liner. Thus, when viewed in isolation, the ethanol areas of replicates are unreproducible, but when an internal standard with the same organic character (Isopropyl alcohol) was added, the RSD dropped to 1%.
Thank you for your attention.

[nipe]

N-propanol - inner standard, 250 mkL, 1:30 split. It's enough.

[MSCHemist]

I do ethanol occasionally. I do total evaporative.

10ul of aqueous sample or if oil/fat based acetone spiked with acetonitrile as an ITSD in a 10ml headspace vial.

120degC incubation for 10 minutes
syringe 125
inject 5:1 split.

I use GC/MS SIM and my quant runs from 10ppm to 1000ppm

[Gizmo]

Glad your problem has been solved. Thought I'd throw out there that we've had some good prelim results going SPME for headspace etoh in aqueous solutions utilizing SIM - ion 31 as quant ion. Could be worth looking into for those without auto HS equipment or otherwise have time to explore an alternate method.

[MSCHemist]

m/z 31 don't you get some interference from oxygen m/z 32? I use 45&46.

[Gizmo]

So far, for our purposes, I can't say 32 has been a factor - although that could change when we move to the validation stage. 45 and 46 were problematic in our system as we were not able to achieve consistency. - also, need to mention that I mistakenly indicated SIM when we actually work in scan. SIM worked in a few trials as well but the sensitivity was not necessary for our purposes.

[MSCHemist]

So far, for our purposes, I can't say 32 has been a factor - although that could change when we move to the validation stage. 45 and 46 were problematic in our system as we were not able to achieve consistency. - also, need to mention that I mistakenly indicated SIM when we actually work in scan. SIM worked in a few trials as well but the sensitivity was not necessary for our purposes.

I wouldn't recommend scanning below 35-40 m/z. That will put extra wear and tear on the detector having all the oxygen being directed to it by the quads. If you don't need the sensitivity I would recommend just using an FID for ethanol. That is what forensic labs use for BAC they use an inlet splitter with 2 columns with different selectivity and 2 FID's. Given what a small molecule ethanol is an FID is probably more sensitive than an MS on scan.

I use 41 for acetonitrile and 45/46 for ethanol and routinely get R2 of .99999-1.00000 and 2-3% RSD's though I am thinking of devoting my old Tekmar 7000-5890-FID to it.

[Gizmo]

I appreciate those points - we will take another look at the larger ions as FID is not currently an option for us