Peak area variations, part 2

[albragi]

Hello,

I am here again with the same question and some clarifications. I posted on Jan 15/2008:

<I experience peak area variations I cannot explain. It is RP chromatography, sodium phosphate monobasic/phosphoric acid buffer (pH=3): acetonitrile (70:30), isocratic for main peaks with gradient later to wash out garbage. SST is always excellent. Samples are bracketed by standards, pattern of injections -- std, two samples (first preparation), two samples (duplicate), std. Two injections from first preparation have the same area between them, two injections from duplicate preparation have the same area between them, but duplicate preparation samples always have smaller area than first preparation (RSD 3 to 10). Sample preparation is simple and no mistakes are made. I even tried to fill those vials (first and second preparations) with a sample from one preparation only -- the result is the same -- different areas. Equilibration of the column seems to be OK, new column makes no difference. The system seems to be OK. Can you help, please?>

And here are some answers to the questions posted:

Tom: area counts are 3700 and 3500 (RSD > 3%).

Uwe Neue: concentrations are the same or almost the same, around 50 ug/ml; carryover is insignificant, 0.2% at the most.

JM: sample is stable; if I re-inject 1st sample preparation after 2nd preparation, peak areas are back to normal.

I should have mentioned that the sample contains approx. 0.6% tween 80 and 0.1% chlorhexidine gluconate, while none of these are in the system suitability solution and standards. Still, I do not understand what is going on. It has to do something with adsorption, but the pattern is strange. Why is it that it happens only after moving to another vial? No evaporation occurs, and all the glassware (vials) and plastic pipettes and filters are rinsed thoroughly before the run. In any case, samples are prepared in the same way. A little hint would be that if I equilibrate the column for 4-5 hours prior to run, RSD decreases, but still stays high. Needle wash solutions tried ACN:water (50-80:50-20), methanol:water (50-80:50-20). Tried without needlewash -- RSD lowered, but still slightly high, the pattern stayed the same.

Al Bragi

[DR]

LC make and model, flow rates, back pressures, any standard trends matching that of samples?

[albragi]

Varian ProStar 230 (pump), autosampler ProStar 410 variable volume, UV detector ProStar 325, flow rate 1.4-1.5 ml/min, back pressure around 70-100 atm (gradient), no standard trends -- standards' responses are excellent (RSD<0.1%).

[AdrianF]

In any investigation it is vital you study the effect of the other components.

I would study the effect of Tween and Chlorhexidine by adding various amounts to the standards. If this turns out to be the reason for the variability I would try extraction using SPE which would remove the other components.

As another possible way round the problem I would try diluting the samples further to lower the amount of tween/chlorohexidine.

[DR]

Agreed - given that the standards are all fine, it appears that it's some sort of matrix effect. Solution: reduce the matrix concentration via any of the mentioned means available to you.

[ym3142]

unlikely, but this can explain: if you injection std1...std4, std5, sample1, sample2, sample3, sample4. carry over from std4 conincidently appear in sample 1; std 5 to sample 2.