strange ionization in LC-MS/MS

[kazy]

I’m trying to develop a LC-MS/MS method for tricyclic antidepressants but I’m encountering major problems with the ionization of the analytes. The concentration-peakheight relationship is not linear, it is exponential. How can that be? And even the peaks of the internal standard (which has the same concentration in each calibrator) get bigger the higher the concentration of the calibrator is.
The working conditions are as follows: I’m using a sunfire 2.1x100mm; 5 µm column. The eluent is 33/67 acetonitrile/5 mM Ammonium acetate + 0,1 % formic acid. The MS is a Quattro Micro operated in ESI+ and I'm meassuring 14 MRMs with 0,2s dwell time. I’m injecting 20 µl sample.
Does anyone has an idea of how I get it working? I so long for a linear calibration curve.
Please help KG

[MG]

I have never used peak heights for calibration, always peak areas. Do you get the same result using areas?

Exponential could mean several things. Do you mean that the slope increases with increasing concentration, or that the slope decreases with increasing concentration (roll-off)? (With concentration on x-axis and response on y-axis).

Is the 0.2s dwell time the total dwell time, or the dwell time per MRM transition (meaning a 2.8s total cycle time)?

How wide are your chromatographic peaks?

What solvent are your samples dissolved in? The reason I ask is that your injection volume is large enough for a 2mm column that you could have mismatch issues, especially in an isocratic method.

[kazy]

yes I'm using peakareas for calibration but it is the same with the heights. and the slope increases with increasing concentration. The peaks get disproportionally big.
I've also tried a dwell time of 0.1s but that makes no difference.
presently the samples are dissolved in water, so there should be no problem. What do you mean by mismatch issues?

kazy

[MG]

Mismatch issues would be if your samples were dissolved in ACN or MeOH, you'd get distorted peaks or split peaks. Since they are in water, that shouldn't be a problem.

And even the peaks of the internal standard (which has the same concentration in each calibrator) get bigger the higher the concentration of the calibrator is.

I missed that the first time. I can think of several possible causes for this.

(1) Your instrument response is drifting upward throughout the run, perhaps due to not being equilibrated long enough, or perhaps there is another problem. To check this, pick a mid-level standard and reinject it several times (5 or more) and see if the response is stable.

(2) A preparation error. To check this, reprepare the curve yourself if someone else prepared it the first time. Also make sure your analyte stock solution isn't contaminated by ISTD.

(3) If your ISTD coelutes with an analyte and is structurally similar, you could be observing signal that appears to come from your ISTD but really comes from the analyte (there are a couple of ways this could happen).

To check #2 or #3, inject a mixture of your analytes that has not been spiked with ISTD. If you still see signal for ISTD, you know that either your analyte stock solution is contaminated with ISTD, or your signal for ISTD is nonspecific and coming from one of your analytes.

[Ericz]

Your MS detector might get saturated. Reduce the injection valume and try again. The linear range can not be very wide in both ends.

[RA Koster]

MG has a point at possible cause no 3.

Does a sample of only one analyte show peaks at the other analytes (if they elute at the same time? Crosstalk

I think your analytes don't elute at the same time, and probably you don't have a deuteric IS?

If that's so, make a calibration curve in academic solution (water) and inject these without analytical column, so you've elimated most parameters, like column and complex matrix.

If this calibration curve is linear, then design a step gradient like beneath:
0 min : 50% aqeous and 50% ACN
2 min : 50% aqeous and 50% ACN
2.1 min : 10% aqeous and 90% ACN
4.5 min : 10% aqeous and 90% ACN
4.6 min : 50% aqeous and 50% ACN
7 min : 50% aqeous and 50% ACN

I have encountered this problem before, and getting to co-elute the peaks was the only way to get a linear curve.

But this will only work if the calibration curve is linear whitout the column!

I used HFBA buffer to trap my component on the column that time.

I hope this will work for you.

[RA Koster]

I forgot to say that in the case i described you will get ion suppression from both components to eachother.

In my case it proved to give perfectly linear results.

I've talked to experts about this and they hadn't got a good scientific explanation.

[kazy]

I did some of the experiments you suggested.
Here are my results:
- My MS seems to be properly equlibrated because I get the same results if I inject the same sample 5 or 6 times.
- I checked the analyte and ISTD solutions but they dont produce crosstalk.
- I also prepared a calibration curve with only one analyte, but I got the same results as before.(the slope increases with increasing concentration).
- The ISTDs I use are deuterated forms of my analytes.
- I dont get better results, if I perform the analsis without the analytical column.

Do you have any ideas as to how to solve the problem?

Thanks to all of you

[bert]

Kazy,

How did you prepare your calibration curves? Is there any sample preparation involved (biological samples) ??

Regards Bert

[kazy]

Hi Bert,

no, there's no sample preparation involved. The analytes are dissolved in water and the calibration curve was done by serial dilution. I've not yet come to the biological sample part.

kazy

[bert]

Did you inject a calibration curve with one analyte, without IS. Is that linear. Is it possible that absorption of your analyte to containers you use can play a role??

Regards Bert

[kazy]

Yeah, I did something like that. I prepared a calibration curve with only one analyte and used an IS of another analyte which has a different retention time (so quasi without IS). It gave the same results. The slope of the curve increased with increasing concentration.

kazy

[bert]

Kazy,

a very long time ago, I worked with tricyclic antidepressants. As I remember well (but again, its a long time ago ) we needed to clean all glass ware with strong acid to prevent sticking to the glass surface. This would be most noticed at lower concentrations................

[kazy]

that means, I should try plastic vials. I'll definitely try if it works.

thanks Kazy

[bert]

Please let us know, and good luck!