LC-MS/MS poor reproducibility of analyte peak areas

[mass50]

Dear Group
I am developing a LC-MS/MS (ESI) method for determining tetracycline in biological samples, but I am facing problems regarding poor reproducibility of analytes peak areas in standard solutions. Relative standard deviation of the internal standard (demeclocycline at a fixed concentration) was about 10% for 18 injections (3 replicates of six calibration point levels injected randomly), but it is now varying systematically arriving 35%. I am using a gradient elution on a 3µm Pursuit C18 analytical column (Varian), dimensions 100x2.0mm, with methanol, acetonitrile and water all containing 0.1% formic acid. The organic mobile phase reaches 70% in the gradient. Before the calibration curve, blank samples (solvent) are always injected in order to stabilize the system. Retention times are far from the dead volume and reproducible. Injection sample composition effect was evaluated and the injection sample composition is weaker than initial mobile phase (20% methanol). Other parameters were checked, such as vial and septum adsorption of analytes and autosampler problems (it was qualified, set at 4°C and methanol acidified with 1% formic acid is being used as rinsing solvent in a needle wash program). Carryover was not observed injecting blank samples after standard solutions at the same analytical conditions. Could the analytes be sticking to any parts of autosampler, to peek tubing or to the column? Why did it work before and does not work now? What kind of rinsing solvent do you suggest? Do you think I have to modify my LC method? Do you have any suggestion of suitable analytical column? Does anyone have same idea of what is happening and how to solve this?
I am getting crazy!
Thanks for help.

[Uwe Neue]

What is the flow rate? What is the gradient run time? The RSDs for standards should be more like 1%, and not 10% to start.

[mass50]

The flow rate is 0.2mL/min and the total gradient run time is 25min.

[Uwe Neue]

Tetracylines are tricky compounds, but I have not heard about a problem of the nature described by you. Commonly, it is difficult to get good peak shapes supposedly due to the complexing of the analytes with metal ions. Commonly citric acid buffers are used, which of course are not suitable for MS.

[Uwe Neue]

Tetracylines are tricky compounds, but I have not heard about a problem of the nature described by you. Commonly, it is difficult to get good peak shapes supposedly due to the complexing of the analytes with metal ions. Commonly citric acid buffers are used, which of course are not suitable for MS.

[mass50]

Thank you for replying, but do you have any other tip?

[Uwe Neue]

here is an application with TFA, which is MS compatible:

http://www.waters.com/WatersDivision/pd ... try168.pdf

[Mary Carson]

I am not familiar with Pursuit. If it is a silica based column, you may be seeing peak degradation due to aging of the column and exposure of active sites. Badly tailing peaks are harder to reproducibly integrate. This is usually obvious if your system suitability includes some kind of tailing test. A new column will solve that problem.

TFA isn't my first choice of modifier, especially if you're going into a mass spect. It may chew up your column faster than formic acid, too. However, I have gotten nice tetracycline peak shapes with it.

Finally, I assume you are making your standards fresh daily from a methanolic or ACN stock solution. Tetracyclines are not that stable in acid--they undergo epimerization and to a less extent dehydration.