strange autosampler behaviour

[Alex Buske]

Hi,

I had issues with inacceptable low precision om one of our RP-Methods running on an Ultimate system. I think i pinned it down to the autosampler.
During sample stability studies, standard solutions and samples were injected every 6 hours from the same vials.
On positions A3 and A4 as well as C4 peak areas went down by about 2%, while on A8 , B7 and D8 peak areas went up byabout 3%. Some of them are double injections with excellent agreement.
The same behaviour was reproducible on another day.
On a second system there was no clear relationship between position and trends, the RSD however were worse.
I reduced draw.speed from 20 to 3 and increased draw.delay from 3000 to 5000 (25 µl injection) and changed the autosampler solvent to mobile phase composition.
On the first HPLC differences decreased, i have however still vial positions that get higher peak areas (b3 and b8 IIRC) on the second system there was not so much change: the deviation were just slightly smaller.
The method is on two APIs and the effect are abserved for both. the tray is cooled down to 5°C.
Any similar obsevations or ideas?

Alex

[HPLCaddict]

Hm, never had issues like that, fortunately. Incredible how the chomatographic world keeps on developing new problems .
Just an idea, as you're using autosampler cooling, could it be that the temperature is not homogeneous over the autosampler rack? So that the actual temperature of the vials depend on the position of each vial? This might lead to density differences of the sample and thus slightly different sample concentrations. I'd try to test the autosampler with cooling off. One variable less...

[paulw]

Agree with HPLCaddict that non uniform temperature could be an issue. It can also be an issue if you put the vials in at room temperature and start the run immediately creating a situation in which some of the later injections had time for the solution in the vial to cool down creating the difference.

I am not familiar with the autosampler you are using. The only similar thing I saw that I pinned on the autosampler was a situations where the line on the injection needle got caught and when moving to certain positions it slightly kinked, so it was drawing slightly less solution than intended at those positions. This did not give me repeatability from multiple injections of the same suspect positions.

I would take the same vial (to eliminate as many sources of error as possible), and make two to three injections out of the same vial from each vial position. Give the vial time to acclimate to any temperature change for the new position and then look at the %RSD of those injections.

[Alex Buske]

Thanks for the input,
I also thought about tmeperature and density. 10° would change the density by about1%. This effect is not strong enough to explain. The vials are placed for 1h in the cooled autosampler prior analysis.
Injections from the same vial at the same time are pretty reproducible.
The overall precision improved when we lowered the draw speed of the syringe. That indicates a "bottleneck" somewhere in the capillary/needle.