Why does my baseline drop?

[Mike H.]

I'm using a synergi Polar RP column 250x4.6, 4µm

Mobile phase A = 20mmol pH 6.5 potassium phosphate dibasic/ MP B= acetonitrile

Gradient is T=0 MPA 100%
T=7 min MPA 100%
T=25 min 40% MPA;60% MeCN

Can anyone explain why the baseline is dropping by so much at ~23min? Thx

[danko]

Because at that point the stuff responsible for the absorbance/baseline rise does not elute anymore, but stays in the column until the initial conditions are re-established.
Actually it wouldn't surprise me if you prepared a mixture of MPA:MPB 40:60% in a test tube it would be rather cloudy one.

Best Regards

[tom jupille]

Another possibility is a lot of "junk" in your sample, in which case a blank would look better.

And a third possibility is accumulated garbage from your A solvent, in which case you would see the problem with both samples and blanks, and the magnitude would increase as a function of equilibration time between injections

[HW Mueller]

I don´t know so much about gradient methods, but this sudden drop is strange. Is it repeatable? What does it look like when you don´t inject anything?

[Mike H.]

I work in a Pharmaceutical environment so we only use Optima grade solvents/reagents Milli-Q water etc. Contamination of aqueous mobile is probably unlikely. I pretty sure the buffer won't crash out here. The drop is reproduceable in samples, and no injection blanks. One weird thing I noticed is that it is not present at >290nm. I thought it might have been some solvent mismatch but it only occurs with this Synergi Polar RP column. When using a XDB phenyl column with same conditions gradient is typical.

I did some googling and found out that it's called a shark fin peak and it seems to only occur with Synergi columns. I e-mailed Phenomenex tech support and was told that it is common with these columns and that decreasing the slope of the gradient minimizes the drop. I e-mailed back asking if they have a mechanism for this.

I really don't want to develop a method with this issue and be forced to explain it to FDA auditors. Any follow up from you guys will be helpful. Thx.

[danko]

The slope you see is not exactly due to actual absorbance but rather the so called “refractive index effect”. I call it that, anyway.
So, as long as there is salt (in this case potassium phosphate) in the efluent you'll observe the slope. But at some point there is too much organic in the system which is the reason for the salt to “choose” to stay on the stationary phase, attracted to the polar groups of this particular stationary phase (a kind of HILIC mode). That is the explanation of the fact that you don't see the same pattern on the Zorbax column – no polar sites there.
The reason that the effect is most pronounced at short wavelengths is the fact that the shorter the wavelength the more refraction is to be observed.
Finally, I find the suggestion to reduce the gradient slope in order to minimize the effect quite funny. Some solution-finders there

Best Regards

[Mike H.]

That's making like 80% sense to be. Let me bounce this back if you don't mind.

At initial conditions the salt is sticking to the polar/ionic site on the stationary phase. When the gradient starts the acetonitrile starts bumping the buffer off of the stationary phase. at some point in the gradient either the salt is completely gone from the SP or the ACN concentration is too high for it to partition to the mobile phase?

So this would imply that to avoid it all together we need to stay below a certain ACN/buffer %?

[danko]

My idea is slightly different:
At initial conditions the salt is going through (not retained) but there is no organic in the mobile phase. You see, the effect I'm referring to is typically observed when there is organic solvent in the eluent – as well as salts. When you begin mixing methanol or can or whatever organic the beam in the flow-cell is refracted a bit and that is “translated” by the detector as absorbance, because not all the light is reaching the detector's PMT. But as described earlier when too much organic is applied the salt (f. ex. the potassium phosphate) prefers the polar groups on the stationary phase over the organic rich eluent and stays there until more water is available in the eluent.

As for a solution, I can think of one but there could be more, depending on other factors: Use a column without embedded polar groups – unless the column here is the only one that works for your separation. If the latter is the case, then it might be a solution to stay below a certain ACN/buffer %. You can easily calculate it by checking the ACN concentration at the point of the baseline drop.

Bets Regrds

[Mike H.]

Update: Phemonenex just sent an e-mail explaining what seems to be going on. Apparently the dip is from column bleed. The phenyl ether linkage has a strong chromaphor and since I'm running just below the max pH 6.5 (max 7.0), it makes sense that it could bleed.

Do you guys agree with this explanation?

[HW Mueller]

Doesn´t make any sense so far. I didn´t understand whether this is not present when there is a blank injected or that there was no blank (and/or pure solvent) injected?

[danko]

If Phenomenex' explanation was credible, then you wouldn't be observing a dip - on the contrary!

best Regards

[tom jupille]

The reason that the effect is most pronounced at short wavelengths is the fact that the shorter the wavelength the more refraction is to be observed.

Actually, in my experience it's the other way around: RI effects are worse at longer wavelength.

If Phenomenex' explanation was credible, then you wouldn't be observing a dip - on the contrary!

Calling the problem a "dip" is something of a misnomer; "shark fin" is more descriptive. It's a return to a "normal" baseline after an abnormally steep rise.

The fact that the problem does not occur in blanks eliminates "A-solvent contamination" as the cause, but also calls into question the "column bleed" hypothesis. One possibility is that it's bleed triggered by something in the sample. When you ran blanks, were they diluent blanks (i.e. same as your actual sample) or were they "mobile phase" blanks (i.e., the initial solvent of your gradient)? Or, if you ran both, did you see any difference.

It's a interesting problem but all in all, the best approach might be to just use a different column.

[danko]

Actually, in my experience it's the other way around: RI effects are worse at longer wavelength.

Sorry Tom, but what I've learnt in school - some years ago - was: The shorter the wavelength the more refraction. If it's not true I'll get my physics teacher for that – hope he's still around.

The fact that the problem does not occur in blanks

But it does occur in the blanks. See what Mike H wrote:

The drop is reproduceable in samples, and no injection blanks

No injection blanks meaning nothing is injected but the gradient is run through – I presume.

Whether we call it a dip (dip, compared to the highest point of the gradient hill) or return to normal level, it doesn't change the fact that the fall does not coincide with the end of the gradient which would be the case if nothing tricky happened.

Aside from all that, I think the bleeding explanation does not make any sense at all.

Best Regards

[HW Mueller]

Well, at sundown/sunrise the short wavelngth are scattered, the long reds come through.
The only time I have ever seen something like this particular "shark fin" is when an air bubble formed in the cell, then was flushed out when it got big enough (isocratic).
Just wonder what happens if the gradient is run without a column.

[ym3142]

This reminds me what happened to me many years ago. I was using some A:B=phosphate : MeOH in an isocratic method. I opened an new Gemini column and set up to stabilize this column. But after many hours (>3hours) I still saw the base line is drifting downward.

I do not want to campaign against Phenomenex. I found I receive more promotion/advertisement than any other column vendors. In the meantime we have more Phenomenex columns in our lab than any other vendors, for which there are two possibilities, or both 1) people trust what they claim in their ads; 2) the column life time is so short that we have to keeping one buying. One of our Gemini methods can kill one column with 200 injections(not always).

Go back to the topic. the "bleeding" explains: the bleeds are not eluted until certain organic strength but the baseline drop when the eluted amount is much larger than the generated amount. This cycles.