Formic acid causing baseline drift in gradient elution?

[kmaudens]

I have a (rather theoretical) question about PhotoDiodeArray detection at wavelengths between 195 nm and 250 nm.

System setup:
Merck-Hitachi LaChrom HPLC system consisting out of a pump with low pressure gradient accessory (type L7100), an autosampler (type L7200), a column oven and a PDA detector (type L7455).

Method setup:
Eluent A consists out of HPLC grade water + 0.1% formic acid (=1000 mL water + 1 mL formic acid).
Eluent B consists out of gradient grade acetonitrile + 0.1% formic acid (=1000 mL acetonitrile + 1 mL formic acid).
The formic acid is Pro Analysi grade (99.8% pure according to C.O.A.)
A linear gradient is programmed from 5% eluent B to 95% eluent B in 30 minutes.
No injection was made.
Oven temperature is 30°C.
PDA detection is made between 195 and 250 nm (slit 2 nm, automatic spectral bandwith, 200 ms interval).
An auto-zero is performed at the beginning of the run.

This analysis results in the following 3D-graph (the scale is in Aborption Units):



At the lowest wavelengths (e.g. 195 nm) there is a steep linear decline of the baseline.
At about 210 nm the baseline is more or less flat.
At higher wavelengths (e.g. 230 nm) there is a steep increase of the baseline.
From about 245 nm the baseline is again flat.

What is the mechanism causing this phenomenon?
Without formic acid in the eluents, the baseline has only a very small increase during the run at all wavelengths (probably due to the UV cut-off of acetonitrile and the change in refractive index during the run) (data not shown).
At first sight, one would think that the formic acid concentration is constant during the run (formic acid in equal concentration in both eluents).
I have read somewhere that the pi-pi interactions from formic acid would change if the environment becomes more organic ... but without detailed explanation or literature reference. Could someone explain this?
Or am I hitting the ball completely wrong and is another phenomenon going on?

Thanks for yours answers and/or suggestions!

Kristof

Note: also relevant references to literature are well appreciated!

[Uwe Neue]

I think you have a multitude of different things going on. For one thing, the absorption of formic acid increases with decreasing wavelength. I am puzzled by the decline of adsorption of lower wavelength, and my opinion is that your detector is just going blind. If somebody knows better, please correct me!

I am also puzzled by the entirely flat baseline at the beginning of the gradient. It should increase with decreasing wavelength due to the absorption of formic acid.

Of course, it is possible that the detector automatically zeros individually all the wavelengths that you are monitoring at the beginning of the gradient. In this case, you are observing just the shifts in detector response as a function of the organic solvent composition at every wavelength individually. I think this makes most sense...

[kmaudens]

The detector indeed automatically zeros individually all the wavelengths at the beginning of the gradient.

I think I found an answer by the book "High-Perfomance gradient elution. The practical application of the linear-solvent-strenth model" from Snyder and Dolan (in my opinion the most extensive, up-to-date and best book concerning gradient elution). In section 5.4.1 (Baseline drift) a "special case of baseline drift" for trifluoroacetic acid in water-acetonitrile gradients is described, with reference to a paper from Winkler et al. (J. Chromatogr., 361, 1986, 191-198 ). Similar phenomena in the same wavelength range are described for trifluoroacetic acid. In short, for TFA, the 200-260 nm region is dominated by the relatively weak n -> pi* transition of the undissociated molecule in acetonitrile, and the strong pi -> pi* band of the anion in the aqueous solution (note: the spectra of 10 mM TFA in acetonitrile and water are quite dissimilar).
Although TFA is of course a much stronger acid than formic acid, I guess the principle could be transferred to formic acid.

But maybe someone is having another opninion or better explanation?