Baseline Step

[JBush]

All,
I have searched through several resources including this forum's search feature and more experienced co-workers and have again returned to Chromforum for help.

I am experiencing a somewhat large "step" in the beginning of my gradient.

Conditions:
Column: Symmetry C18, 4.6x75mm, 3.5um column
Detection: 225nm
MPA = 0.1% Formic acid in water
MPB = 0.08% Formic acid in Acetonitrile
Gradient: T = 30C Flow = 1.5ml/min
time %B
0 5
10 95
11 95
11.1 5
15 5

The step comes at just before 2 minutes and the void volume is approximately 0.65minutes. I have come to the assumption that it is merely a mixing issue and can not be avoided, however it would be comforting to know for sure that there is no way of eliminating this step with the above or similar conditions. Note that a similar method on a longer column seems to near eliminate this step.

[skunked_once]

Try preparing 5% acetonitrile as your mobile phase A then your starting conditions can be 0% B which will avoid the mixing issue if that is the cause. You can also make mobile phase B with 95% acetonitrile then your gradient can be 0-100% over 10 minutes.

[Mark Tracy]

I have seen the step you talk about when using similar conditions. What is happening: The system equilibrates at isocratic conditions of 5% MeCN. A certain amount of UV-absorbing "stuff" is adsorbed on the entire column. Then you start a steep MeCN gradient. The "stuff" desorbs from the column and shows as a step. It is later than void because the new equilibrium lags behind the solvent front, which is a constantly changing gradient.

This desorption effect is much worse for strong "ion-pairing" agents like TFA. Also steep gradients make it worse. Finally, the "stuff" may not be the formic acid itself, but some minor impurity in your water.

This desorption effect also has the effect of magnifying mixing noise, famously a problem for TFA-containing mobile phases. Depending on the design of your LC gradient pump, you may find mixing is not so good at low percentages.

[JBush]

Mark, This desporption effect you speak of makes sense as I am near the lower limit of the pump's abilities to provide the exact amount of mobile phase B (5% +/- 1 or so) It would also explain the odd baseline ripple/spike I am seeing within two minutes (roughly) following my gradient. However, I will try the 5%ACN in MPA as Skunked Once has suggested. The thought had occurred to me but was halted by my supervisor earlier this week. I have some time today to try it and will let you guys know if this helps. Thanks Again!

[JBush]

For those interested, starting my mobile phase A with 5% ACN and running a 0 to 95% organic gradient with the previous mobile phase B did NOT eliminate this baseline step. It was worth a shot and I appreciate the thoughts!

[JA]

JBush,

Does the baseline step occur at the same time when using the long column as it does with the 75 mm column? If it was something being desorbed by the gradient I would expect it to take longer to show.

Do you see the step in blank and sample runs? How about if you use a ZDV union in place of the column? Is there a step if you run isocratically?

[tom jupille]

If it's an artifact generated by the onset of the gradient, you would expect it to come out at (gradient delay time + column dead time).

[Uwe Neue]

Tom's suggestion triggered another thought: what happens if you prolong the preequilibration of the column with mobile phase A?

[JBush]

Using a longer column (150mm in place of the 75mm) all other dimensions identical the step occurs at approximately the same time. This step is in both blank diluent runs as well as system blank runs (no injection). There is no step in an isocratic run and I have not tried running the gradient program with a ZDV.

Uwe Neue, when you say preequilibrate the column do you mean a 1 or 2 minute isocratic before the onset of the gradient? As far as equilibration goes I have added plenty of time after the gradient run and have tried the conditions I have just mentioned. Added time before hte gradient does shift the step by about that amount of time.

My concern is when we start looking at degradents we may run into a problem? Again thanks for all the help!

[Uwe Neue]

If I understand you correctly, it has nothing to do with a column preequilibration. It is something that occurs at a particular point in the gradient run time, but it is not related to the solvent composition at the start of the gradient. This is puzzling.
I suggest to run a blank without the column in place. Then, with the same setup, measure the gradient delay of your system.

[JBush]

There is no step when running the method with a ZDV union in place of the column. A change in column type(Symmetry C18 to YMC ODS-AQ) did not eliminate the step while a large increase in gradient run time (10 min to 30 min) reduced it by about half height. It looks to me as if without changing mobile phases/column dimensions I may be stuck with it. Also changing UV does not eliminate it but at much higher wavelegth it becomes a step down instead of up, the baseline looks flatter but the signal drops significantly. Just wanted to follow up on where I'm at.

[Mark Tracy]

That sounds exactly like what I have seen also. I also observed that the height of the step is proportional to the acid concentration. I even tried using a ternary gradient program with the acid on a separate channel; same step. I finally decided to live with it.

When I did my study of acid additives, the only ones that did not cause a step in the baseline were phosphoric and perchloric.

[tom jupille]

Okay, the fact that it's wavelength dependent and can go negative suggests that it's a refractive index anomaly due to an abrupt change in mobile phase composition.

The fact that you don't see it without the column suggests that it is not a mixing problem at the pump, but is tied in with column chemistry.

This sounds a lot like the "demixing" phenomenon that I used to run into three decades ago doing thin layer chromatography. In effect the stationary phase preferentially sorbs the stronger component of the mobile phase so that the advancing solvent (or gradient in this case) gradually becomes a "step" gradient. I think that this is what was on Uwe's mind when he asked about the column pre-equilibration time.

The catch is that this explanation is not consistent with the elution time of the step being independent of column dimensions, so the whole thing still leaves me puzzled.

One last thought: if it is an RI anomaly, it would be worse on a detector whose flow cell is out of alignment with the optical bench. If you see the same problem on several instruments, there is probably not much you can do. If you see one instrument that's worse than others, it's probably time to have the detector serviced.