Weird trouble with an optic cell (Agilent 1260 DAD, G4212)

[vadzim]

Short description with pictures:
https://imgur.com/a/2R6A1tC

Longer description:
Agilent 1260, DAD G4212
Pretty straighforward program: HPLC water with pH 2.5, no salts - Methanol gradient:
99:1 (hold for 2 min) change to 20:80 over 9 min, return to initial composition.
Regular baseline has a slight elevation (or 'bump') in the middle.
During the analisys gradual shift of baseline occurs toward slight downward slope in the middle. Then during one of the runs a sudden 'jump' of the baseline, and on the next chromatogram, where initially was a little 'bump' due to gradient, is now a huge drop (better see the pictures). The baseline profile stays about the same after this. Turns out the problem is with the optic cell. Changing the cell to another solves the problem, and strange downward profile reproduced with the problematic cell on another HPLC.
The problem is not just weird baseline form, it is also more noisy and affects minor analyzed components.
We tried to purge the problematic cell with HPLC water heated at 60 C, both forward and backward flow - it didn't help. Any ideas as to what could be the problem and what can be done about it?
Thank you.

[Multidimensional]

Lots of basic info missing (flow rate, column dimensions, comp program etc) so no way to provide any firm diagnosis. HOWEVER, it looks like you may have purchased one of their detectors that uses the flow cell refereed to as the: "Max-Light Cartridge Cell". If so, this may be why your flow cell needed to be switched out. In our experience (3-decades), the max-light flow cells appear to be extremely fragile and either easily fracture (leak) or clog (partially of fully obstructed) resulting a range of strange signal outputs, especially as the mobile phase composition changes OR the system back-pressure changes (such as when someone runs MeOH/Water gradients where the pressure max's at 50% composition then drops either above or below that %. These cells are not designed to handle more than ~ 800 psi (check documents/manuals for exact value) and if any type of clog appears, even for an instant, the cell quickly over-pressures causing the silica tubing (cell) inside to be damaged (changing the light path). As the composition changes, so does the light path and strange peaks or baseline changes sometimes are observed during use. Replacement of the flow cell always cures the problem 100%.

The Number One Problem Part: In our experience, we see all of the 60mm path length 'Max-Light Cartridge Cell's FAIL in this way. They have the worst reliability over the other cells due to the extremely long path length silica fiber.We have been to so many labs over the past years where clients had drawers filled with broken 60mm cells (and they are EXPENSIVE) due to damaging them on a regular basis. We strongly recommend that no one purchase them at all (they have DADs that use the std cells, avoiding the problem). We see problems with the 10mm Max-Light cells too, but far fewer of them overall. While it may or may not be your issue, please be aware of this issue and next time you purchase a DAD try and buy the one that only uses the STD spring clip style flow cells (not the optical fiber version). The spring clip flow cells with quartz windows are extremely rugged and rarely ever fail; even after abuse. They have a built-in over-pressure feature which allows them to leak, the reseal when the pressure drops (these have been around for 30 years).

[DR]

I did not know about the max light issue, though I know Thermo had somewhat similar issues with their longer Surveyor light pipe flow cells.

If it is not the flow cell, it could be your matrix.
Having a pH 2.5 aqueous A phase absent salts means it's probably also absent buffering capacity. If your matrix has more buffering capacity than your buffer and it's not a match for your standards, you're going to have problems, especially if your pH effectively wanders too close to an analyte's pKa due to lack of buffering capacity in your mobile phases.

[TylerSmith123]

Hello there,

Take my advice with a grain of salt because it seems like you have tested other detectors and they work fine in this system. However, if I'm reading your software correctly, are you observing your targets at a wavelength of 223 nm? If so, this may be a matrix issue like DR is saying. Depending on your additive, the region between 200-225 nm can be muddied with your matrix absorbing a lot of that wavelength. This is typically an issue for people who use additives such as formic acid. Anyway, it really depends on the accuracy of your measuring here and the purity of your phases. The less accurate you are in making your mobile phase, the larger differences you will see as two phases switch between each other, while you will always see a less-than-ideal solvent grade absorb more at those lower wavelengths.

[Hollow]

As Multidimensional said, some more info about the chromatography would be needed (column dimensions, flow rate, full gradient programm and run time) .

When you write: "change to 20:80 over 9 min", means that the 20:80 mixture is at 9 min (absolut) or at 11 min (2 min of 99:1 + 9 min for the change)?
And what is you gradient delay (dwell volume)?

So what comes to my mind (ignoring that changing the flow cell helped. Probably something else changed too, like batch of mobile phase, sample? (Water quality, cleaniness of glasware?)):

a) do you have any late eluters that just start to elute at your final 20:80 and is not fully out of the column when you start re-equilibrating?
(This may also be specific to the batch of mobile phase or contaminated glasware, if the method normaly runs just fine.)
How long do you keep running the final composition?
Maybe you can increase your final methanol concentration to about 95% and keep it for 1-2 column volumes (before going back to initial and re-equilibrate for about 5 column volumes + 1x your gradient delay volume) and see, if something comes out.
If this is just sitting in the flow cell when it autozeros for the next injection, I would expect strange baselines for sure.

b) cleaning the flow cell with just warm water may not be enough if this is something lipophilic. The "magic mix" referred by Waters in a former version of their "Controlling Contamination in UPLC/MS and HPLC/MS Systems Guide (715001307)" and their KnowledgeBase (WKB238676), may be a nice try and helped others in former posts.
Just mix equal volumes of Water, Methanol, Acetonitril and Isopropanol and add about 0.1-1% of formic acid and let it run e.g. overnight at low flow rate.

[vadzim]

Thank you everyone for your time.
Great ideas and considerations. Sorry for not providing more info about chromatographic conditions. In my head they were clearly not the problem, so I dismissed the particularities. Just to give some closure: pretty standard conditions - 150x4.6, 5 um, Zorbax SB C18, 1 ml/min, 25 C, gradient as described above. Sample is a water soluble drug product, main ingredients - sucrose, paracetamol, ascorbic acid, aspartame, and a flavor.

Very insightful info from Multidimensional. Indeed the cell in question is G4212-60008 - Max-Light Cartridge Cell 10mm. And this is the third time we have had a problem with that type of cell over the course of about 7 years on 4 HPLC systems. It seems, it’s time to get a new cell. :/

[vadzim]

A bit of follow-up:
Washing with hot water for HPLC (60 C) didn't help, washing with i-PrOH however solved the problem, probably, quite literally. We pumped through the cell about 200 ml of i-PrOH at about 0.5 ml/min. As far as I can tell the issue have disappeared. Maybe this will be helpful for someone else.

[DR]

I gather you're taste masking acetaminophen in either a chewable or ODT formulation.

IPA is probably removing an excipient (polymer) residue. These formulations are a major PITA for the analytical people left to deal with them. They're even worse when natural flavors are used.

If gradient baselines start going wonky with blank injections, check your degasser...

[Multidimensional]

If washing the flow path with " i-PrOH" helped to restore the signal, then: (1) Your current Sample prep and/or the injection solution may not be suitable for this application. Something may be accumulating or precipitating out; (2) The fiber was probably clogged with something that is obviously soluble in i-PrOH. The method used is not properly dissolving the sample material leading to precipitation of material in the flow path. A std flow cell can deal with this issue. The "Max-Light Cartridge may not. The detector's fused silica fibers are certainly one of the areas it is likely to be captured, perhaps resulting in a partial obstruction (as noted in my earlier post). These partial obstructions result in changes to the flow cell's light path; (3) Maybe a better wash method needs to be used that follows good chromatography fundamentals (the wash solution should follow the analysis (not be part of it!) and contain a solution that is stronger then the mobile phase and which dissolves all retained material 100%.