Chromatography Forum

Hi everyone !

My question is simple : Is it possible that an analyte could have interaction with the filter used before putting in vials ?

The authorities recommand to justify that the filter used, doesn't interact with the analyte.

So to justify it, we have to compare a non-filtered sample to a filter one ! But I think it's not good to inject a sample without filtration !!!

(I talk about a pharmaceutical sample ; for example with APAP)

In my opinion, we have to choose the good filter regarding the compatibility with the eluant.

What's your opinion ?

Thks

Make one sample, filter aliquots through every commonly available filter you can get your hands on...

Or you can make your sample centrifuged,and inject liquid supernatant, then compare peak area of centrifuged and filtered sample.

One can do this sort of thing with standards first, these don´t need filtration.

Or you could do a little theoretical research or get an appropriate catalog from some supplier who sells filters made from different materials. In these catalogs it is convenient for the manufacturer to give lists of solvents/chemicals that are compatible and/or incompatible with the chemistry of the filter.

Regards

If I remember correctly - a typical filter study in our lab was:

bracket std (used for quantitation)
sample 1 first mL aliquot into vial
sample 2 (discard 1 mL) - 2nd mL aloquot into vial
sample 3 (discard 2 mL) - 3rd mL aloquot into vial
ect. (maybe up to 5 or so)
bracket std

The final method always stated "Discard 1st 2 mL aliquot", ect.

I'm lucky that I don't work in an environment that is very strictly controlled, and few of my samples are very dirty. I used to filter everything, but increasingly my clients started asking exactly these sorts of questions. Since I was handling a very wide range of samples, all very different to one another, it became very time-consuming assessing each one.

In the end, I realised that my C18 guard columns were costing less than £20 each (I work in the UK; sorry, I don't know the conversion to dollars). This meant that I could afford to block a guard column every 20 runs and still come out cheaper than some brands of spin filter!

Nowadays I don't filter anything, but I spin everything, and change guard columns whenever pressure is bad. Even so, I don't want to admit how many runs I get from a guard column, because in proper QA environments, you would probably be replacing your analytical column more often than I replace my guards!

The loss of analytes on filtes is well known and some results have been published (search J. Chrom. Sci. around 2000). The losses depended on the analyte, concentration, solvent, and filter type. In short, just about anything is possible.

Bryan's suggestion is best. Check the first few mL from the filter to see if there are any differences.