Ghost Peaks & Bad Gradient Baseline

[DR3Chemist]

We have 2 identical Agilent 1200 series HPLC's with a Quat Pump and VWD detector. Each is running the same method with the same column and mobile phase sources. Here is a brief description of the column, mobile phase, and method conditions:

Column: Agilent Eclipse XDB-C18
Mobile Phase A: 0.05% H3PO4 in Water
Mobile Phase B: Acetonitrile
Run time: 80min
Gradient Profile: 0-25min 66% A / 34% B
25-55min ramp to 10% A / 90% B
55-70min 10% A / 90% B
70-80min 66% A / 34% B

Problem: One instrument recently started giving ghost peaks and baseline weirdness on the ramp portion of the run and then lines out semi-clean during the 10/90 portion. The first analysis after the LC is started up for the day (after sitting overnight with 66/34 in the system) looks terrible (see image below). Subsequent runs are not near as bad. We found that if we flush the system with Acetonitrile for 10-15min during startup, the first blank analysis will be better but still not acceptable. Our 2nd LC with the same conditions continues to run fine, so I have been working on the assumption that something is contaminated in the problematic LC. During troubleshooting, I have tried many different things to clear this up, but it will not go away. Here is a list of what I have tried (not necessarily in order and somethings probably would not cause my problem, but I am grasping at straws):

- New mobile phases
- Mobile phase from a different water source
- Different brand of acetonitrile (all LC quality)
- Different H3PO4
- Flushed system with IPA
- Flushed degasser channels
- Swapped degasser
- Swapped MCGV (had a new spare)
- Replaced AIV cartridge
- Swapped AIV (had a new spare)
- Swapped pump module from another LC (not from the sister system to this one)
- Bypassed autosampler
- Bypassed column compartment

The first problem source that I went to was mobile phase, and I still think it has something to do with this. I do not have the knowledge to logically explain why this occurs. The longer the LC sits with the pump turned off, the worse the baseline will be during that first analysis. I think we could ease the problem by running a flush at startup, but we have ran this method on these 2 LC's for almost 15 years now (with regular preventative maintenance) without needing a startup flush.

I'm at the end of my limited expertise in solving this. If anyone has any thoughts or suggestions to share, I would greatly appreciate it.

Normal Blank Analysis


Problem Blank

[TylerSmith123]

Hello DR3,

I may not know exactly what is causing your issue, but I could give you some more places to check. The first thing I would suggest checking is maybe your filter frits in the mobile-phase itself if you think it could be originating from the mobile phases. Usually, during PM, they will replace these along with a slew of other things, but who knows. This may also be a case for the pump head seals/o-rings, and potentially the pump-head washing solution. I know that you said you used a new pump in place of your old one so that is probably not the issue. Clearly though, you have narrowed this problem down extensively and isolated a variety of different parts of your LC without any improvement, so it must be something that you have yet to try.
When you are running samples, does the signal ever increase? Ie, if you were to EQ for a long period of time would this result in large contamination peaks during your "wash" portion of the method?

Sorry I couldn't be of more help, but hopefully this will invite other potential solutions to your problem

[DR]

What I'd do:
I'd assume that somewhere in the system you've got a little ecosystem because your A side is 100% aqueous. I'd passivate the system, then I'd quit using 0.05% H3PO4 as your A phase.

Premix A and B MPs to match your starting and ending mobile phases, then run 100% A to 100% B with whatever slopes you want.
This gets enough ACN into your A line, frit etc. so that you're no longer growing anything in the system. It may also provide you with an improved baseline.

If this is a dedicated system for the method mentioned, never stop flow. Lower flow to a few percent of what it is when you're testing and shut the lamp off between runs, but keep the pump going - it might help minimize the issue as well.

[JI2002]

Assuming the system was working well before, what changed before the ghost peaks were observed? Were the ghost peaks increasing gradually over time or did they appear suddenly?

Based on the answers to the questions, it may require different approaches to figure out how to troubleshoot the problem.

I generally don’t replace instrument parts before finding out the source of the ghost peaks:

1. Replace the column with the good column from the second (good) instrument to rule out the column;
2. Inject air ( empty vial) to rule out the solvent/sample;
3. Bypassing the autosampler by connecting the pump directly to the column to rule out the autosampler;
4. Replacing the mobile phases with the mobile phases from the second (good) instrument to rule out mobile phases.

Hopefully the source of the ghost peaks can be identified after the steps above. If not, focusing on the modules before the column, especially the pump (autosampler has been ruled out at this point).

[DR3Chemist]

Thank you guys for the responses, they are much appreciated.

@Tyler: The ghost peaks never get worse with subsequent runs. The first run through after cranking up the LC (and running for about an hour on 66A/34B) looks like the "problem blank" above. Subsequent runs will still show some smaller peaks that match up with the larger ones in the "problem blank". If we flush the system with acetonitrile at start up, the following runs will be like the runs after the problem blank...smaller ghost peaks, but still too large to accept.

@DR: I also suspected a small algal bloom at one point and hoped the IPA flushes would have cured any such presence. We also replaced the glass frits for the mobile phase reservoirs. These instruments run a compendial pharmaceutical method, so I have little wiggle room in the method parameters. Adjusting the mobile phase as you suggested is well within those parameters, so I will give that a shot next week when I get back. We could easily pre-blend the mobile phase and go from there. We have traditionally simplified the makeup of most of our gradient mobile phases for ease of prep.

Leaving the LC pumping on a lower rate in between runs is an option, but one we would rather not do. This instrument runs for about 4hrs out of 8, 24/7/365, and we usually turn the pump off in between runs to save mobile phase. It has only been getting left off overnight lately during troubleshooting. Even on the regular shutdown timing, we see baseline upset and ghost peaks. Another option would be to try leaving 10A/90B sitting in the system overnight and see how that goes.

@JI2002: The system had ran great the past few months leading up, but this all started during the same time we got a bad batch of Acetonitrile from our supplier. I didn't mention this before because all of our other LCs cleared up immediately after swapping out the bad ACN. This one mostly cleared up but had some lingering ghost peaks that have since gotten worse. The system has been cleaned, flushed, bypassed, and swapped out so much since then, I don't see how that could still carry over.

[TylerSmith123]

Hi DR3,

What exactly was the issue with the acetonitrile? I wonder if you have something stuck in your system that is slowly leaching out over time, but you should have probably been able to isolate the part responsible already if this was the case. Could this issue be originating from the sinker frits of your mobile-phase? How long was the instrument running the troublesome phase?

[Mike H.]

80 min run time? This seems long even for a 250mm column. I doubt 0.05% H3PO4 is going to be generating algae growth. I've had 0.1% H3PO4 on LC for over a month and never seen a problem.

You might be picking up something from the bottles? To isolate if it's a MP problem, run at initial condition conditions for 20min and inject. If peaks get bigger, it's from aqueous MP.