Polymyxine B sulphate variable retention time

[aceto_81]

Doesn't the method require the sample solvent to be 80:20 H2O:CH3CN?. I'd also try dissolving the sample in the mobile phase itself.

I tried this yesterday: here are my retention times for one peak:
inj H2O/ACNMobile phase
1 16.002 15.578
2 15.917 15.829
3 15.775 15.507
4 15.673 15.966
5 15.876 15.862
6 15.405 15.532
Remark: my peaks for my samples dissolved in mobile phase were doubling at the end of my sample set (from injection 5). Immediately after these samples, I began to inject the samples dissolved in H2O/ACN. The retention time increased immediately from 15.532 to 16.00 and my peak shape is back to normal.

My suspicion is that the repeated injections of samples is affecting retention.
I'd try injecting varying numbers of mobile phase blanks between samples and seeing if the retention time variations change.

After seeing above numbers, its hard to see if the variations change I think.

I'd also email the EP experts, and ask them for suggestions. The database also suggests a Waters Symmetry column, and I'd try that.

Please keep having fun,

Bruce Hamilton

I will send the EP experts an email. I haven't tried the Symmetry yet, but after trying 3 different columns (ABZ+, BEH Shield RP18 and a BEH C18), I think we can say that it isn't a column problem?

Please keep posting,

Ace

[HW Mueller]

What are "my peaks"? Polymixine + corticoid or only polymixine? Thus: What was splitting (doubling)?
Are the polymixine peaks quite broad, fairly noisy at the apex (you may have to expand the chromatogram to see every data point to check whether a noisy apex influences rt)? Does the start and end of the polymixine peaks shift parallel to the rt?
How did you obtain the mobile phase used in dissolving sample?

[aceto_81]

What are "my peaks"? Polymixine + corticoid or only polymixine? Thus: What was splitting (doubling)?
Are the polymixine peaks quite broad, fairly noisy at the apex (you may have to expand the chromatogram to see every data point to check whether a noisy apex influences rt)? Does the start and end of the polymixine peaks shift parallel to the rt?
How did you obtain the mobile phase used in dissolving sample?

my peaks is only polymyxin, corticoid is ok.
The splitting/doubling I meant was first the development of a shoulder (3 injection) which evolutes to a second peak (5th injection).

the peaks are quite broad, but not noisy.
In my previous runs, with the ACN/H2O as solvent, the start and end of the polymyxine peaks are shifting parallel, there was no peak distortion.

I obtained the mobile phase by making a fresh solution of 4.46g/l sodium sulphate brought at pH 2.3 and then added 200ml ACN to 800ml of the solution. My mobile phase was a solution, made up the day before. For the moment I'm running with exact the same composition (made up my samples with premixed mobile phase, the same bottle as is put on the UPLC).

Ace

[HW Mueller]

Did you really look at the peak apex under the expansion I suggested?
Anyway, the rt of broad peaks varies a bit more than that of sharper peaks. Primarily it looks like a pH and/or ionic strength fluctuation and resulting partial incomptibility is the underlying cause. If you discipline your act a bit more the rt fluctuation should be reduced.

[Bruce Hamilton]

Danko,

"Once you've" is future tense, not past or present.

Ace,

I'm a little confused, you imply you previously dissolved the samples in 80:20 H2O:CH3CN, but in an earlier post you suggest sample solvent is water.

Also do you obtain the same variation for both CRS and your product?.
I'd be reluctant to change compendial methods unless I'm sure they don't work on the CRS as well as my sample. I assume the original analysts managed to achieve acceptable results.

The appearance of a split peak may indicate a solvation issue, so I'd stay with the method 80:20 H2O:CH3CN sample solvent, but reduce concentration to half ( 0.25 mg/ml instead of 0.5 mg/ml ). Could make variation better or worse, but it's worth a try.

Anyway, looking at your numbers, aside from the first sample, there are no obvious trends, so I'm with Hans, being very precise in all aspects ( including trying a premixed and degassed mobile phase ) may help improve precision.

Uwe is the best person to comment on whether trying a Symmetry column could have any effect on precision.

Please keep having fun,

Bruce Hamilton

[aceto_81]

@ HW Mueller: Yes, I really took the time to take a look.
@ Bruce Hamilton: as soon as the discussion began about the solvent strength, I made the switch to 80:20 H2O:CH3CN. I get the same variation for CRS and product.

Yesterday I put up some runs with a double concentration of sodium sulphate, and my variation is acceptable, (with mobile phase as a solvent, but also with water).
Afterwards I put up the normal concentration to get a comparison, and the variation was still acceptable (only tested with mobile phase as a solvent). (Everything was premixed)
I flushed the column afterwards with water and CH3CN, so now I'm going to inject again with the normal concentration, to see if there was an effect from the previously pumping with double sodium sulphate concentration.

For the moment I think I'm going to keep the things like they are now: premixed mobile phase and working very precise and then the variation should be fairly limited (at least this is what I hope).

Thanks to everyone who has posted, I learned a lot.
If one still think for improvements, please post!

Thanks


Ace

[danko]

Hi Bruce,

Danko,
"Once you've" is future tense, not past or present.

Yeah, sorry for the misunderstanding. It’s because I focused more on the phrase:

I'd be less keen to continue to play with buffer strength…..

The thing is; I really think the issue here is related to the nature and the ion strength of the mobile phase. Ace is partly dealing with HIC interactions, but due to the length of the mechanism description, I chose to jump directly to the potential solution.

Ace wrote: For the moment I think I'm going to keep the things like they are now: premixed mobile phase and working very precise and then the variation should be fairly limited (at least this is what I hope).

Thanks to everyone who has posted, I learned a lot.
If one still think for improvements, please post!

You might like to get back and tell us how the next results are (once you’ve acquired them) i.e. whether everything’s fine or the problems reoccur.

Best Regards