%RSD problems

[RDRR]

Good morning/afternoon all,

Working on a headspace method here for the analysis of ethanol in a pharmaceutical suspension. Samples are prepared in de-ionized water, with isobutanol as an internal standard. GC is a Shimadzu 2014 GC with an CTC autosampler (syringe) and FID detection.

The problem we're encountering is %RSD failing for linearity and accuracy preparations. Criteria is set at 10% RSD, which is pretty large, yet %RSD values are all over the place.

Linearity is from 2.5 ug/mL to 1000 ug/mL. Vials for each level are prepared individually, with internal std added (125 ug/mL isobutanol). For linearity, we do 3 injections of each level.

Accuracy samples are prepped a bit differently. Suspension is measured out (1000 mg) into a 10-mL volumetric flask, and spiked with ethanol to concentrations of 250, 500 or 750 ug/mL. From these spiked solutions multiple vials are prepared with internal std and injected once each.

GC conditions as follows:
Column: Zebron ZB-624 column, 0.32 mm x 30 m, 1.8 um layer of phase
Carrier gas: He @ 35 cm/sec
Split ratio: 1:10
Injector Port Temp: 110 C
Detector Temp: 250 C
Column Temp Program: 40 C for 1 min, ramp to 80 C @ 15C/min, ramp to 240 C @ 50C/min, hold 240 C for 1 min
Injection volume: 1.0 mL headspace

Headspace conditions:
Incubation Temp: 80 C
Incubation Time: 7.5 min
Syringe Temp: 105 C
Agitator Speed: 300 rpm

Any suggestions are most welcome. I've already tried changing liners, septa and even replacing the injecting syringe.

[Ron]

There are a couple of things I would suggest to try. First, the most common internal standard used for ethanol is n-propanol. I would suggest the trying the propanol, the properties are more similar to ethanol. The incubation temperature is high for this analysis. An incubation temperature of 40 to 60C is more common for this analysis. Your equilibration time may be too short for proper equilbration. You could try a 15 minute equilibration time to see if this helps.

On the GC end of things, you don't need any glass wool in the injection port liner, the sample is already in the vapor phase. I would also raise the injection port temperature to around 200C.

For the CTC autosampler, I would make sure the injection port depth during injection is set to at least 45 mm. Are you using an inert gas flush of the syringe after injection? This can be very important when determining polar analytes in an aqueous matrix to eliminate carryover.

[krickos]

Hi

Agree with Ron and will expand a bit.

Why lower HS oven temp?
When using water as sample solvent you should stay below the boiling point of the solvents of intrest to obtain a stable equilibrum. This is not necessary at the same extent when using DMF for instance as sample solvent.

Minimum HS equilibration time, are some different data in litterature but one basic concept is the minimum time that is depended on sample volume and is a trade off for sensitivity. In some cases you may do with 10mins if your sample volume is like 1ml in a 10ml vial or 2ml in a 20-22ml vial but it might be a close call.

So by lowring HS oven to 60-70°C and extending equilibration time to 15mins you likely will get a more stable and repeateble equilibrum.'

As Ron stated, no wool in liner and it should be a low volume one like 200ul or so to minimise peak broadning and possible interactions in liner.

Another potential issue later with accuracy. You mentioned that samples are suspensions. How will standards be prepared? External or standard addition? You may encounter a sample matrix issue if external standards are used depending on whats in the suspension that can cause high or low recoveries. Polar residual solvents in water is particular sensitive to sample matrix effects.

[Consumer Products Guy]

First, the most common internal standard used for ethanol is n-propanol.

Agree. That's why I couldn't myself understand why <USP 611> utilizes acetonitrile as internal standard for ethanol assay until I realized three things:
U
S
P

[Ron]

I wish USP would go to a consensus type of method approval process like ASTM does, and I never thought I would look at the ASTM process as better than any other approval process. There just doesn't seem to be anyone at USP doing a thorough technical review of the methods to see if they make sense and are at least as good as commonly accepted methods for the same analytes in other industries.

[qcChemist]

I work with the USP on a daily basis, and we are always finding holes in their methods. For that headspace analysis, I would also suggest that you keep your transfer line temperature at least 10-15 degrees above your equilibration temp, as there will be some heat loss during your injection.

[AICMM]

RDRR,

I agree with Ron about raising the injection port temp. I would also ask what the RSD for butanol is, all by itself? It seems this should be fairly reproducible (more so than ethanol) which should give an insight into the problem. I would also ask what the injection speed is? Right now you have about 10 cc/min into the inlet, maybe 20cc/min?

Best regards.

[chromatographer1]

My guess (no information on vial size or sample size was given) is that

you may have poor sealing on your vial septa.

This is the single most likely cause of high RSD values.

The second most common cause is supersaturation of the headspace.

This is caused by too much sample. Ethanol equilibrates in 8 minutes at 80°C if the sample solution volume (water, DMF, etc) is less than 250µL. You should have RSD values of less than 2% for ethanol when the amount of ethanol is less than 100µg in the headspace vial. I often got less than 1%.

I believe Peter Apps in So Africa achieved even lower RSD values in his research.

If your solution is at 20g/L (a 2% solution - 20µg/µL) and you use 250µL then you have a sample weight of 5000µg in solution. If you spike ethanol at 100µg the residual impurity level is 2% in the sample (100 / 5000)

Headspace is inherently more accurate than the best technician can prepare samples. "At least it is when it is performed properly," he said, after being humbled repeatedly in the lab.

When you prepare your samples in the proper dynamic range of your analytes and vial - sample volume you will see extraordinary results.

Good luck in your work.

Rodney George
consultant

[Ron]

I must say I am kicking myself right now, I overlooked something very basic and obvious. Since you have the CTC autosampler, set it up for liquid injections and do a series of liquid injections of the same samples and see what the RSDs of the samples are. If RSDs are good the problem is somewhere in the headspace sample introduction, if the RSDs are the same you may have a GC problem. I would raise the injection port temperature before doing this and have glass wool in the liner.

[Peter Apps]

All that has gone before is good advice.

With butanol as an internal standard in aqueous ethanol with a Gerstel clone of the CTC and an Agilent 6890 I got rsds of around 0.8%, but ONLY if I had the sample oven at room temperature. The temperature control on the ovens is poor, and the thermal inertia is not high enough to stop the temperature fluctuating while a sample is withdrawn or a new vial is added to the queue.

Injection rate was also quite critical since you want to avoid pressure pulses inthe inlet that lead to eratic split ratios.

Peter