sudden drop in baseline

[jzt]

We have a strange issue with the baseline during method development. The compounds of interest are very polar (piperazine derivative). We screened 7 columns that their manufacturer claim to be good for polar compounds. We obtained reasonable separation on the Synergi Polar RP column (4.6x150 mm) with the following conditions:
Mobile phase A: 0.1% TFA in water
Mobile phase B: 0.1% TFA in acetonitrile
Flow: 1.5 mL/min
Gradient: 5% B to 40%B in 5 minutes, then to 90%B at 15 minute.

We observed a sudden drop in baseline (signal goes from 150 mAU to about 20 mAU in 0.2 minute, on Agilent HPLC system). After that, the baseline drifted down slowly while the %B was increasing (from 20 mAU to negative 30 mAU in 7 minutes). This happened on different columns (same type), and on different HPLC systems. It shows up in both blank and sample injections. And it doesn't matter if we use the DAD (reference on or off) or VWD detector. When the gradient is changed, the time when this drop happens will shift but always around the same %B. We have an impurity peak that sits right before that “cliffâ€

[jzt]

Thanks to tips from SYX, here's my attempt to post the chromatograms.

[img][[img]http://img105.exs.cx/img105/3017/baselinedrop5wz.jpg[/img]

JZT[/img]

[Kostas Petritis]

jzt,

What do you get if you do the same gradient without a column? How confident are you that the TFA is still good (i.e. do you use 1 mL TFA vials)? What happens if you do not use TFA in your mobile phase (not in terms of separation quality but in terms of your baseline).

[Uwe Neue]

Do you get a similar blib with all columns?

[tom jupille]

What wavelength? and does the size of the blip change as a function of column equilibration time (pre-gradient)?

TFA forms a charge transfer complex with ACN, and the actual absorbance spectrum of TFA in ACN/water mixtures changes as a function of ACN concentration. If memory serves, there's an isosbestic point on the spectra at 214nm (an isosbestic point is sort of a "pivot" at which absorbance stays constant as conditions change). The equilibrium distribution of TFA between the mobile phase and the stationary phase surface is also dependent on the ACN concentration.

I'll throw out a hypothesis that what you're seeing is the combined effect of TFA being adsorbed/desorbed by the column and the spectral change. If that's true, changing to 214 nm should help (unless you're already there!) as should fiddling with the TFA concentration in the A and the B solvents. I've seen separations where people use, for example, 0.1% in A and 0.08% in B with the explanation that "that seems to give a better baseline".

[Rob Burgess]

There is very in-depth discussion on baseline disturbances in the following ref.: K. Choikhet, B. Glatz and G. Rozing; LC-GC Europe, Feb 2003 issue. I couldn't make much sense of the srticle but it does describe how you can get some strange baseline affects when using TFA/AcN gradients.

We have seen problems with baselines with TFA gradients and the use of 1 mL ampoules does decrease the baseline "noise". However I've never seen any baseline disturbance to the extent you are seeing. Sorry I can't be more help. Good luck!

[jzt]

Thanks to everyone who reply to this post. I should add a few details.

This happened with several Synergi columns. We used other columns on the same HPLC and didn't see such big drop in baseline. We are using the 1 mL TFA ampoules to prepare mobile phases. The detection wavelength is 215 nm. However, when we were using a DAD detector, we also looked at 254 nm, 230 nm, and 220 nm. the baseline drop were evident at all these wavelengthes to a different extend. This does not apprear to related to column equilibration time. We are aware of various problem that may araise from mobile phases containing TFA. However, our sample matrix has TFA. so for simplicity in sample preparation, we stayed with TFA mobile phase. the suggestion that "0.1% in A and 0.08% in B seems to give a better baseline" is puzzling but we may give it a try.

Thanks,

JZT

[tom jupille]

I don't know that 0.08% TFA in B was a global optimum; try changing that percentage and see if it affects the peak.

Also, let us know what happens so we can all learn!