Impurities with new or old column

[GT]

Dear all,
I have this big problem:
I analyze HPLC purity of my API with reverse phase HPLC, using a Symmetry C18 column with water and acetonitrile as eluent.
Using a "new" column (from 0 to about 100 injections) I can see only 2 or 3 impurities peaks. If I use an "old" column (more than 1 or 2 hundreds injections) I found on my chromatograms (of the same sample) new impurities peaks with no neglet value (near or above 0.1% of main peak).
What do you think about this ?

[tom jupille]

If it's the same sample and the same column, one possibility is that your sample has degraded during the period of time that it took you to run the 100+ injections.

If it's the same sample on the same day, but comparing two columns (new and older) then the column chemistry has changed. This is not unusual. In particular: if either your API or any of the degradants have ionizable groups and you run without a buffer in your mobile phase, then very small changes on column chemistry can make surprisingly big changes in selectivity..

[GT]

Dear Tom,
thanks for your help.
The situation is the second you wrote: same sample, in the same day analize with 2 different column (new vs old). But the next question is:
which is the real purity of my sample ? Can I think that extra peaks (obtained with an old column) are ghost peaks or degrade in column ? Otherwise are they really impurities of my API and I have to use old column to quantify them ?

[tom jupille]

Given that it's the same sample, the results from the column showing more peaks is probably closer to the truth.

The fact is that chromatography is always a negative technique as far as qualitative analysis goes: if you have one peak, you have at least one compound, but there is no way to guarantee that there are not additional compounds co-eluting. For that reason, many labs will develop "orthoganal" chromatographic methods (methods with the largest possible differences in selectivity) to provide the maximum chance of seeing all impurities.

Note that selectivity differences or changes can cut both ways: a change may separate formerly unresolved peaks but may also reduced or eliminate separation of formerly resolved peaks.