Column Lifetime

[The_Hamster]

We are currently writing an SOP on column performance (i.e. parameters for determining when a column should be discarded). We have acceptance criteria for resolution and symmetry of peaks eluting from the columns in question when injected with a column check standard. We are wondering if other people have guidelines they use for elution time shifts when determining if a column needs to be discarded. We are considering the resolution and peak shape to be the major determining factors but would like to say something about % change in retention time as well.

The Hamsters

[Mark Tracy]

If you want to set a spec for retention time, 1) use an isocratic test 2) prepare the mobile phase gravimetrically, 3) use a column thermostat. As to how tight the spec should be, that depends on how often you can recalibrate the retention windows on your methods. Your intuition is right, don't put too much weight on this test. Critical pair resolution, peak symmetry, and selectivity are more important.

[danko]

Hi

I agree with you both, that resolution and tailing are the 2 most important indicators/SST. Maybe plate count would be nice – especially if you don’t have closely eluting peaks to benchmark.
What I disagree with is the need for separate SOP for a column or potentially columns.
Firstly you’ll have to conduct the tests in a different setup then the routine daily analytical runs, because you’ll have to change mobile phase and/or other parameters. And most importantly the results of these tests won’t be directly linkable to the analytical method you’re using the column/s for.
So if I were you, I’d include these tests (resolution, symmetry or whatever is critical) in my method of analysis and set some acceptance criteria e.g. Resolution >= 1.5, Symmetry 0.9 – 1.5 and so on.
Then I’d start every analytical run with these tests (SST) and the day the criteria are not fulfilled is the day I’ll have to do something about (regenerate, dispose of, or whatever) the particular column. Provided the problem isn’t due to the HPLC equipment.

Best Regards

[pliva]

We are currently writing an SOP on column performance (i.e. parameters for determining when a column should be discarded). We have acceptance criteria for resolution and symmetry of peaks eluting from the columns in question when injected with a column check standard. We are wondering if other people have guidelines they use for elution time shifts when determining if a column needs to be discarded. We are considering the resolution and peak shape to be the major determining factors but would like to say something about % change in retention time as well.

The Hamsters

dear hamster,
we do have sop for rp column testing but its a big trouble because according to our sop we have two approaches to test it. first one is our in house method and second one is the method used by the vendor. the acceptance criteria is we should get at least 70% of what the vendor claimed. the prblem arises when the column fails according to our inhouse method. the vendors method need to inject 0.5µl and our system do not have that facility as its PQ is up to only 5µl only. so we need to have the system which the vendor have.
according to my opinion, i think we should avoid this procedures for testing the column because during validation also we have proved that the result is not affected while using at least two different column. its ok when we test the column it gives theoretical plate say 6000 but its not necessary that while routine analysis it shoud give 6000 because the analyte is different. i feel we should not used the column when it fails system suitability. as long as our selectivity does not have any problem i think its ok. but nowadays most of the cro started this practice and it seems to be overcautious.

[koen_shimadzu]

I would do it in the following way.

1. Create a SOP for RP columns in case 1 column is used for more applications. Especially if there is much deviation in pH and column temperature. Don’t look to the vendor protocol, but dig into the literature for selecting a protocol.
2. If a column only is used with one method (which I would prefer), set some acceptance criteria. Be aware that in case you’re results are out of spec, the system is out of spec and not only the column. It can also be the mobile phase, column temperature or system causing the problem .

[Bruce Hamilton]

Consider these:-
- If my employer buys new columns to replace columns that are still working OK, am I compromising the chance of a decent pay increase?.
- If I don't replace columns that give incorrect results, will I still have a job?.

For new columns, on arrival, I attempt to replicate the manufacturer's CoA on my HPLC, so that establishs baseline date. It's really important to ensure that column connections have no dead volume, and please don't use a guard for column performance testing. Often, I see about 90% of plates, but it has ranged from 70 - 110%.

For unique columns, this isn't a pass-fail test, as even the calculation methods used can vary the results, but if I purchase several columns, I have a feel for typical values on my system. In my experience, columns from major manufacturers are pretty consistent, but smaller, specialist column producers let some dodgy columns through.

If I use a column for a single type of analysis, I try to establish acceptable performance criteria from historical data. If I don't have any data available, and the method is critical, I ensure that I have a new column as backup. I scrutinise the chromatogram to determine critical resolution and peak shape criteria, and monitor those.

If the column is for general purpose use, I replicate the initial CoA data, and define a limit, usually based on 70% of the initial column plate count that I obtain, and/or peak shape, and regularly monitor performance, depending on use.

The fun thing about monitoring column performance is that I discover very quickly when I've made major oops, and created a small void or coated the column with rubbish.

When I discover voids on columns for research and development, I usually just pump the column up to 2-3 times normal flow for 15 minutes to compact the bed, and fill the voids with material from the back end of used guards, repump up to high pressure, and then retest.
Usually gives me another 3-6 months of column life.

If you want to try and fill voids, please ensure you know the type chromatograms that voids generate, so you only open columns that have voids, otherwise you could total a good column. At best, your chance of a pay increase will diminish.....

Bruce Hamilton