Reproducibility problem

[robbertvw]

Hello all,

I have a strange problem going on which I would like to have you're opinion on.
The past two weeks I have been doing an Itraconazole analysis and from the beginning I'm having problems with the reproducibility. When doing 10 injections from the same reference solution I'm for example getting 4 injections with a peak area of 4600 mAU and 6 injections of 4400 mAU. The RT's are pretty stable.

The analysis is based on a monograph and has been validated internally. The following conditions are used:

System: Agilent 1100 with DAD
Column1: Polaris 10cm, 4.6mm, 3 um with guard column
Column2: Symmetry 10cm, 4.6mm, 3.5 um with guard column
Mobile phase A: Tetrabutylammoniumhydrogensulphate buffer (54 grams per 2 liter) (ion-paired)
Mobile phase B: CH3CN
Gradient elution
Injection volume: 5 ul
Detection: 225nm +/- 16nm

Because at first I thought it was an injector problem I replaced the following items in different stages: rotor seal, stator face, needle, needle seat and this didn't solve the problem. As a final resort I changed to another Agilent 1100 system.
Because at some point I got split peaks with the Polaris column I had to change it to a new one but this kind of column wasn't available and so my supervisor wanted me to change to the Symmetry column. I'm getting the same problem on that column. These columns are reserved for this analysis because of the ion-paired mobile phase.

A week before I got these problems I ran the same analysis just fine. Only thing I can think of now is that we are using a different jar of the buffer salt.

Anybody got any idea's?

[Gerhard Kratz]

Your problem is the buffer and the equilibration time for your column. Ionpairing makes such problems.
Monographs are most time challenging. Who did the validation???? Say thank you to this person.
Take it easy and develop your own, new method with a volatile buffer. Both columns are ok. Good luck.

[ATAlR]

I think that the problem is that you went well beyond the dynamic range of DAD detector which is 2000mA. Try to dilute your sample and try again.

[Consumer Products Guy]

I'd stay under 1000 mAU. USP <621> states injecting less than volume in monographs does not need re-validation.

[HPLCaddict]

Guys, the OP was talking about peak AREA. That one may well be beyond the magic 1000 and still be in the dynamic range of the detector... I'd take a look at the validation data, linearity should have been proven there going even higher than 100% concentration.

@Gerhard: I share your negative opinion concerning ion-pairing, especially if gradients are involved. But in this case I don't think that this is a matter of insufficient (re-)equillibration. Retention times are pretty stable according to the OP. I would expect more variance reagrding retention time than peak area if equillibration is not enough.

First thing I'd do would be some kind of OQ of the autosampler. Throw out the column, use a ZDV union or a pice of low-diameter SS tubing to create some backpressure, use a simple mobile phase (such as pure methanol) and do multiple injections of a simple analyte (such one of the famous parabens). If the autosampler is faulty, you will realize pretty soon. As long as injection accuracy is better than, let's say 1% RSD for n=6, there's no need to swap out more parts of the autosampler and you may take a closer look at the method. If it's worse, than the method is not the culprit.

Did I understand correctly that the very same method runs flawlessly on a different machine?

[Consumer Products Guy]

Guys, the OP was talking about peak AREA.

We don't "know" that for sure, because Area is not expressed in mAU (see below). So robertvw used wrong term for either area or for units, so we took a guess.

I'm for example getting 4 injections with a peak area of 4600 mAU and 6 injections of 4400 mAU.

1. Running system suitability on the autosampler to isolate whether autosampler is the issue - that's a good idea.
2. Robertvw didn't mention replacing the 1100 autosampler metering seal (same part as 1100 pump seal)
3. USP monograph methods - in my mind - are funky way more than straightforward.

[robbertvw]

Ok guys, I have an update:

Yesterday I received a new batch of the Tetrabutylammoniumhydroxidesulphate and I did 10 injections of my reference solution. This time we ordered extra pure quality of this salt instead of normal quality 99% vs 98%.
Now the RSD of those 10 injections is 0.2% and the baseline is way more stable.

[Consumer Products Guy]

Ok guys, I have an update:

Yesterday I received a new batch of the Tetrabutylammoniumhydroxidesulphate and I did 10 injections of my reference solution. This time we ordered extra pure quality of this salt instead of normal quality 99% vs 98%.
Now the RSD of those 10 injections is 0.2% and the baseline is way more stable.

I'm not sure that I can bring forth a reason for that alone helping, as even a mobile phase of higher background should be consistent. But don't argue with success.

[DR]

It may well have been the salt. I was going to ask whether you had filtered your A phase before use. If not, you probably should - any MP that uses pretty much any sulfate as an ion pairing agent.