HS-GC-FID Ethanol Method: bias

[crimgal]

Hi all,

I'm attempting to validate a new method on two Shimadzu GC2030s coupled with their Nexus HS20 for ethanol determination in blood.

Right now, I'm just shooting an aqueous NIST traceable certified reference material 40 times (our runs never exceed anything beyond that). I noticed that as the run progresses, the bias drifts higher. I've done experiments in the past to demonstrate that 10 minutes is ample time for equilibration.

I'm more than happy to share the current headspace and GC parameters but I wanted to see if anybody else experienced this during their validation.

Thanks

[Peter Apps]

Welcome to the forum

Drift during a batch is common with equilibrium headapce analysis (and with other injection techniques).

Nobody can tell you anything useful without knowing the scale of the problem and the details of the method.

Peter

[Consumer Products Guy]

I'm attempting to validate a new method on two Shimadzu GC2030s coupled with their Nexus HS20 for ethanol determination in blood.

I'm going to assume that you work in a forensics related field. That said, I will also assume that you are not really validating a new method, as such would be tough-to-impossible to enter into evidence in a judicial process, as a defense attorney would have a field day with that.

So it appears that you are trying to verify new or other equipment for suitability with an existing validated ethanol method - correct?

[crimgal]

Welcome to the forum

Drift during a batch is common with equilibrium headapce analysis (and with other injection techniques).

Nobody can tell you anything useful without knowing the scale of the problem and the details of the method.

Peter

Thank you for your response, Mr. Apps. Samples that are in the latter portion of the run experience a positive bias nearly 7%. (e.g. a NIST traceable CRM of 300mg/ml quants as 320mg/ml). Any of your insights would be greatly appreciated. Please let me know if there's another parameter that you're interested in, I'll be more than happy to provide.

Internal Standard (I.S.): n-propanol (0.5 v/v%)
Headspace vials: Restek 20ml, septa PTFE/silicone
Samples are prepped at a 1:10 dilution--100ul of sample, 1000ul of I.S.

Calibration model specs: currently on a 6-point calibration curve, working range is 10mg/ml up to 400mg/ml. Quadratic curve, no forcing, no weight.

Instrument A

Headspace
Oven Temp. : 70.0 C
Sample Line Temp. : 150.0 C
Transfer Line Temp. : 160.0 C
Shaking Level : 3
Multi Injection Count : 1
Pressurize Gas Pressure : 14.7 psi (nitrogen)
Equilibrating Time : 10.00 min
Pressurizing Time : 0.50 min
Pressure Equilib. Time : 0.10 min
Load Time : 0.50 min
Load Equilib. Time : 0.10 min
Injection Time : 0.50 min
Needle Flush Time : 0.50 min
GC Cycle Time : 4.60 min

Injector/GC
Injection Mode : Split
Carrier Gas : H2
Flow Control Mode : Linear Velocity
Pressure : 7.0 psi
Total Flow : 52.4 mL/min
Column Flow : 3.15 mL/min
Linear Velocity : 55.0 cm/sec
Purge Flow : 2.0 mL/min
Split Ratio : 15.0

Isothermal program, 35.0 C
RTX-BAC Plus 1 column

FID
Temperature : 300.0 C
Sampling Rate : 200 msec
Stop Time : 4.00 min
Delay Time : 0.00 min
Subtract Detector : None
Makeup Gas : N2
Makeup Flow : 24.0 mL/min
Detector Flow Const Mode : No
H2 Flow : 32.0 mL/min
Air Flow : 200.0 mL/min


Instrument B

Headspace
Oven Temp. : 70.0 C
Sample Line Temp. : 150.0 C
Transfer Line Temp. : 160.0 C
Shaking Level : 3
Multi Injection Count : 1
Pressurize Gas Pressure : 14.7 psi
Equilibrating Time : 10.00 min
Pressurizing Time : 0.50 min
Pressure Equilib. Time : 0.10 min
Load Time : 0.50 min
Load Equilib. Time : 0.10 min
Injection Time : 0.50 min
Needle Flush Time : 3.00 min
GC Cycle Time : 4.60 min

Injector/GC
Injection Mode : Split
Carrier Gas : H2
Flow Control Mode : Linear Velocity
Pressure : 5.1 psi
Total Flow : 36.2 mL/min
Column Flow : 2.26 mL/min
Linear Velocity : 41.2 cm/sec
Purge Flow : 0.0 mL/min
Split Ratio : 15.0

Isothermal program, 35.0 C
RTX-BAC Plus 2 column

FID
Temperature : 300.0 C
Sampling Rate : 200 msec
Stop Time : 4.00 min
Delay Time : 0.00 min
Subtract Detector : None
Makeup Gas : N2
Makeup Flow : 24.0 mL/min
Detector Flow Const Mode : No
H2 Flow : 32.0 mL/min
Air Flow : 200.0 mL/min

[Peter Apps]

Do the peak areas of the ethanol and the propanol increases as the batch proceeds?

Is the increase in ethanol result smooth from the begining of the batch (i.e. how does the plot of ethanol vs vial number look)? Similarly for the ethanl and propanol peaks.

If you increase the enumber of vials in the batch, does the upward trend continue?

Peter

[AICMM]

Crimgal,

Apologies for not having the time at the moment to go into your post at depth yet, but two things stand out to me. First, IS is at 5000 ppm and top of the curve is 400 ppm so small fluctuation in area of the IS would not necessarily be as obvious as fluctuation in ETOH. Second, quadratic curve for an FID in the range you are talking about is really - well - weird. FID should have plenty of linearity in that dynamic range so you should not need quadratic.

Peter's comments about areas over time is very pertinent to my first point above.

I need to think about sitting out versus heating time since they may sit on the tray for a while but they actually sit in an oven for only a fixed, limited time.

Best regards,

AICMM

[crimgal]

Crimgal,

Apologies for not having the time at the moment to go into your post at depth yet, but two things stand out to me. First, IS is at 5000 ppm and top of the curve is 400 ppm so small fluctuation in area of the IS would not necessarily be as obvious as fluctuation in ETOH. Second, quadratic curve for an FID in the range you are talking about is really - well - weird. FID should have plenty of linearity in that dynamic range so you should not need quadratic.

Peter's comments about areas over time is very pertinent to my first point above.

I need to think about sitting out versus heating time since they may sit on the tray for a while but they actually sit in an oven for only a fixed, limited time.

Best regards,

AICMM

The IS should be 500 ppm; sorry, I must have transcribed the IS concentration incorrectly in my original post. For what it's worth, I experienced the same issues with IS concentration half that.

Linear (no weight) could have been selected; the standardized residuals analysis was nearly identical when I compared the linear calibration (non-weighted) model with the quadratic (non-weighted) model. However, the quadratic "won" when I looked at the CRM quants. Linear tended to overestimate very low CRMs (20mg/ml) by nearly 50% which doesn't seem terrible (the difference between 20 vs 23mg/ml seems like nothing), but that's unacceptable bias in my world.

I do agree with you, though. I was honestly expecting the calibration model to be linear as well.

[crimgal]

Do the peak areas of the ethanol and the propanol increases as the batch proceeds?

Is the increase in ethanol result smooth from the begining of the batch (i.e. how does the plot of ethanol vs vial number look)? Similarly for the ethanl and propanol peaks.

If you increase the enumber of vials in the batch, does the upward trend continue?

Peter

Over time, the area counts for ethanol increase as the run progresses. n-propanol does as well, but not as dramatically. When I look at the ethanol/n-propanol area count ratio, I can see it increases from 0.55 (1st sample), and by 40th sample the ratio is at ~0.58.

Let me know if you'd like me to provide you with exact area counts for both instruments.

For what it's worth, I did a run where I only increased one headspace setting: the equilibration time. I doubled it to 20 minutes but saw the same issue.

[crimgal]

Do the peak areas of the ethanol and the propanol increases as the batch proceeds?

Is the increase in ethanol result smooth from the begining of the batch (i.e. how does the plot of ethanol vs vial number look)? Similarly for the ethanl and propanol peaks.

If you increase the enumber of vials in the batch, does the upward trend continue?

Peter

The hyperlink is a screenshot of some charts for the 10-minute equilibration experiment

https://photos.app.goo.gl/vyVG38wpLFTJecULA

[Peter Apps]

The plots of peak area vs vial are really informative - for both peaks you have two reasonably flat lines separated by a step at 29-30 vials (most obviously on instrument A). So there is something that the headspacer is doing to the second 30 vials tha it does slightly differently to the first 30. Or are you, perhaps, filling vials in batches of 30?

Peter

[crimgal]

The plots of peak area vs vial are really informative - for both peaks you have two reasonably flat lines separated by a step at 29-30 vials (most obviously on instrument A). So there is something that the headspacer is doing to the second 30 vials tha it does slightly differently to the first 30. Or are you, perhaps, filling vials in batches of 30?

Peter

As far as I know, the instrument software isn't coded to process the samples differently after vial 30.

All samples were aliquotted continuously (I didn't get up to take a break, etc.) so they were treated all the same.

[Peter Apps]

Maybe nothing to do with the software - from pictures on the web it looks as if 30 vials are queued for analysis - that could easily have something to do with it.

[AICMM]

Two questions. From the pictures on the web it looks like the Nexus only holds 30 vials???

Second, have yo measured the platten temperature over time? Is it poosible this is a contributor to sample heat?

Food for thought is all.

Best regards,

AICMM

[AICMM]

Re: 30, I just realized that is what Peter had already asked.

Best regards,

AICMM

[crimgal]

Two questions. From the pictures on the web it looks like the Nexus only holds 30 vials???

Second, have yo measured the platten temperature over time? Is it poosible this is a contributor to sample heat?

Food for thought is all.

Best regards,

AICMM

The Nexus holds 90 samples. Forgive me for the silly question for you, but what is a platten?