Bad peak shape after x injections

[artsjeroen]

We're analyzing Amoxicillin using a isocratic system.
After about 40 injection the peak height decreases and a shoulder appears on the right side of the peak. Replacing the guard fixes the problem but is not a workable solution.
A dozen columns have been used only this year, so the problem is defenitily not column-batch related.

Any ideas on the reason for the bad peak shape after ±40 injections?

Flow = 1,5 ml/min
Injection volume = 10 ul
Column = C18; 4,6 x 250 mm; 5um (Grace Alltima)
T column = 30°C
Eluent = 50 mM KH2PO4 / MeOH (95:5) pH=5

The concentration of amoxicillin is about 0,05 mg/ml in water.

[tom jupille]

First off, why is replacing the guard not a workable solution? That's what guard columns are for.

I can think of two possible explanations:
1. You have a bad batch of guard columns which are developing a void space in use.
2. There is some sort of CRUD* in your sample which is contaminating the guards.

If you really want to tell (and have the time to spend!) do 40 blank injections and then a sample. If it's a void space problem, it should show up whether or not you injected samples. If it's contamination, then that injection should look OK.

If it *is* contamination, then the only solutions are to continue changing the guards regularly or to modify the sample prep to get rid of the CRUD before injection.

*CRUD = Chromatographically Retained Undesirable Debris

[artsjeroen]

We have ca. 100 injections per analysis. Alas, replacing the guard during each analysis is therefore not workable. And this analysis is performed almost every day.

I doubt it is a debris problem, since we also see this problem using the amoxicillin raw material dissolved in purified water and diluted in purified water. So there is 50 mg in 100 ml water, diluting 5 ml to 50 ml in water. In my opinion this is not a sample that would cause debris.

If you really want to tell (and have the time to spend!) do 40 blank injections and then a sample. If it's a void space problem, it should show up whether or not you injected samples. If it's contamination, then that injection should look OK.

- hmm I could try this yes, unfortunately the 40 injection is an estimation, sometimes the peak is still good after a couple analysis (± 200 injections)

[tom jupille]

In my opinion this is not a sample that would cause debris.

If that's the case, it points toward bad guard cartridges. If the samples are clean, then a guard should be unnecessary. You might want to risk a column and try running without the guard to see what happens.

[Consumer Products Guy]

We're analyzing Amoxicillin using a isocratic system.
Eluent = 50 mM KH2PO4 / MeOH (95:5)

OK, isocratic and elution with 5% organic.

Worth trying:

1. You may need to initiate a cleanout high in organic after a certain number of injections.

2. You may want to experiment with same isocratic elution, then run the organic up higher to clean out the column after your peaks have eluted.

In both cases, you'd need to allow column equilibration time. My guess is that with either of the above potential solutions that you would NOT need to revalidade, as your elution parameters have not changed.

[artsjeroen]

Thanks for the tips!

I've been doing some overwork this weekend.
Changed the eluent to Eluent = 50 mM KH2PO4 pH=5 / AcN (95:5)


Problem is solved, but I can not explain the reason?

[Coffo]

Not sure if this is the case here, but there are examples of methanol adducts forming with amoxicillin ... methanol can attack the beta-lactam ring. Amoxicillin can also form dimers (since it has both acid and base groups present). Can't think why either of these would lead to guard columns failling after a number of injections though - you could always remove the inlet end fitting of a failed guard column to look if there is a void that has formed. If the guard has voided, that is the source of the poor peak shape.
Are you preparing all of the samples at the same time (before the analysis commences) and only seeing the poor peak shape after a number of injections have been made ?
If so, the shoulder could be a result of a slow deterioration or change in the samples waiting analysis (perhaps gradual dimer formation ?); the guard column may not be the problem.
Thankfully it sounds like you've found a cure anyway ...