We have problems with our new HPLC equipment. The DAD baseline decreases or increases at various wavelengths when gradient with increasing acetonitril concentration is applied.This is particularly observed in visible range. The detector uses both deuterium and tungsten lamp. We are interested in analysing writing inks, therefore the detection at visible wavelengths is desirable. We have no problems in our older Agilent system when using the same mobile phase, column and gradient. The manufacturer claims that this is due to different constructions of the flow cell. Our older system contains only deuterium lamp and the drift at visible range is negligible. It is possible to subtract blank analysis in the new system, but it would be much more preferable to get rid of the baseline drift.
/Jan

We have noticed more energy at visible wavelengths with a DAD using the visible lamp (better signal to noise) in the visible range than obtained with same everything but using an Agilent 1050 VWD in the visible range.

when you set up the detector parameters, what are the sample wavelength/bandwidth and reference wavelength/bandwidth ? Since MeCN doenot absorb at the higher WLs, you should look at the purity of the solvents.

We are using several different wavelengths, for example 580 nm with the bandwith of 40 nm for blue/violett inks. If reference wavelength is used at for rexample 700 nm, the baseline drift detected at 580 nm is lower. In our older equipment (Agilent 1090) no reference wavelength is neccessary, the baseline is stable for every wavelength (naturally with the exception for very short wavelengths, where MeCN shows some absorption). We use the same column, mobile phase and method in both systems, but only the new system exhibits the baseline drift when acetonitril gradient is applied . For some wavelengths this drift is upwards, for other in the opposite direction.

Can You rule out that there`s a problem with traces of dissolved gas in the mobile phase? Even if You use the same eluents with another system this could be a problem of the special detector in use.

Two possibilities come to mind: UV mismatch or RI mismatch.

If it's UV mismatch, the drift should approximately track the gradient profile (i.e., it should be linear for a linear gradient). It should also be more pronounced at shorter wavelengths than at longer wavelengths. Usually (but not always), the drift will be positive at wavelengths lower than your reference negative at wavelengths above your referece (be careful about generalizing here, if you have a buffer in the A solvent, it can go the other way!). Running UV spectra of your solvents should tell you what's happening in this case.

If it's RI mismatch, then it will usually not track the gradient profile (RI is a non-linear function of composition for most reversed-phase type mobile phases); for a wide-range gradient, you can actually see a "mid-gradient hump" where the baseline maxes out. It will probably be more pronounced at wavelengths above your reference, and probably be negative below your reference wavelength.

The amount of RI-mismatch drift is a function of cell design, but it is also exacerbated by misalignment of the flow cell. If possble, beg or borrow another of the newer detectors and see if the problem is about the same as on yours. If it is, then it's a design issue. If one is significantly worse than the other, then it's probably a service issue.

Thank you for your comments. The baseline drift should not be due to UV mismatch. I am using 20% MeCN and 80% water with 10mM perklorate, pH=3 as the mobile phase A. Pure MeCN is the mobile phase B. The gradient is linear from 30% to 100% B in 15 minutes. The mobile phase exhibits absorption only at very short UV wavelengths. It can be difficult to chose reference wavelength as the inks absorb both in UV and in visible spectrum, we are normally not using any reference wavelength for ink analysis in the Agilent 1090 system. So the design of our flow cell is probably the reason for the baseline drift.
Concerning the degasing, this new HPLC system uses vacuum degasing, the older system uses He degasing.
Jan