Hello,

We are dealing with a negative peak that appears at a constant retention time in a Waters HP-SEC system. We have three detectors, a UVA, fluorescence, and a total organic carbon analyzer. The negative dip doesn't appear in the UVA or fluorescence detectors. It's only an issue in the carbon analyzer. We have been looking at this exhaustively. A colleague brought my attention to negative dips in her ion chromatography system. They call these the 'water peak' or the 'water dip', and someone working with me also wants to call the negative dip in the TOC signal a 'water peak'. However I am resisting that I think these dips have different origins and to call them the same thing is confusing the issue.

I understand that the volume of water injected creates a mixing zone when it is inserted in the eluent stream, but the different properties of that injected water and the eluent are going to show up in different ways in different detectors, based on how the detectors work. The IC detector is based on conductivity, while the TOC analyzer is based on the wet oxidation of C in the stream and the detection of CO2.

The problem with the negative dip is that it appears at a critical retention time. Perhaps it can be delayed by slowing the flow rate?

I could go on, but I will see if anyone responds to this much.

Thanks for your thought,

David
Civil Engineering and Geological Sciences
University of Notre Dame

I meant to add that the size of the negative dip scales with the volume of injected water, and the peak occurs even with injections of pure water into a pure water eluent.

Dave

The negative dips are also referred to as "system peaks". They are much more common with bulk property detectors (e.g., RI or conductivity) than with selective detectors (e.g., UV, fluorescence). They arise from the fact that a deficit of a component which is normally present in the mobile phase will chromatograph to the same retention time as an excess of that same component.

Not having worked with TOC analyzers as LC detectors, I'll ask a couple of dumb questions:

1. Is the detector response flow or pressure sensitive? (e.g., can you see pump pulsations on the baseline)? If it is, then you may be seeing the pressure surge caused by the injection process. If you can simply cycle the injection valve (do not load any sample at all) and still see the negative dip, then you will have confirmed the problem.

2. If your mobile phase is pure water (i.e., no carbon), how can the signal go negative? (if the output *is* flow sensitive, that would be the answer)? The only other answer I can see is if you have more dissolved CO2 in the pure-water obile phase (is the detector that sensitive?) than in the pure water your are injections.

And one question which may be more to the point:

3. Does the elution volume of the negative dip correspond with the total permeation volume of your system? Unless it's a well-characterized sample, you shouldn't be trying to quantitate at that point in any case; the specificity is somewhere between horrible and nonexistent.

If the problem results from a pressure surge, then changing the flow rate *may* help (I'm not terribly optimistic, though).

If the negative dip is at total-permeation, then you're probably stuck, but you shouldn't be quantitating there anyway.

In principle, dissolving the sample in the actual mobile phase should get rid of the problem (barring issues like trace dissolved gases, per above).

Actually, a negative dip with a TOC analyzer is more easily explained than with conventional HPLC detectors. As Tom said this type of thing results when the "slug" of sample you inject has a deficit of something that is present in the mobile phase. OK so what might be in the mobile phase that is giving a background response? What does a TOC analyzer respond to? Let me think. Oh yeah - EVERYTHING!! At least anything with oxidizable carbon (which is as close to everything as you can get). So if your sample is diluted in very pure water, it is not at all surprising that you would see a negative dip at the void volume, with a TOC analyzer.

I think Tom's suggestion to dissolve your sample in the same thing being used a mobile phase is a good one. And certainly it's better not to work with peaks that elute near the void volume: but it can work in some cases (if all other components of interest have some retention).

Good luck!

Gentlemen,

I didn't want to go into our thinking on this problem of negative dips in my first message for two reasons; 1) didn't want to leave too long of a message right off the bat and 2) didn't want to sway anyone's intuition with my impression on this.

The feeling I have, based on loads of troubleshooting work on the HPLC-TOC interface, is that there is a fairly high amount of some organic contaminant coming through our HPLC. We have independent TOC analyzers here and I have used them to check i) our water, i.e. the water from the reservoir going into the system, ii) the efflent from the system when the column is not in place and iii) the effluent when the column is in place. Our purified water is less than 0.5 ppm in carbon, and the effluent of the system is 3-4 ppm. We have two columns; one column adds ~5 ppm to the effluent, but a new column we are using only seems to add ~ 1.5 ppm.

I believe that the negative dip in the TOC detector is due to the injection volume dependent dilution of this background C contamination.

Two approaches would be to i) get rid of the carbon contamination so it can't be diluted and lead to negative dips or ii) cause that injected water front to elute later, i.e. resolve it better from our peaks of interest. However it needs to come quite a bit later. We are degrading molecules and are interested in seeing the lower MW components that follow on the tail of the main peak.

We have a new column and are going through the process of determining optimal conditions for it, perhaps we are just going to have to start making very long runs when we use the TOC detector.

Ideally I would like to eliminate the C contamination from the system. Then presumably the problem would be fixed. The other strikes me as basically being a bandaid fix. However, I agree it may be very hard to get rid of enough of the bleed.

To respond to a suggestion, yes we did consider the possibility that the issue was originating in a bicarbonate concentration difference between the sample and eluent. We discounted this however by making injections with the degasser off and sparging the sample, and observing no change. Also, we typically observe the C level swing by several ppm during the negative dip, and the bicarbonate level is in the sub-ppm level.

Does anyone have experience they can share in either i) reducing the carbon bleed coming out of an HPLC system, and recognizing points within the HPLC that carbon can enter or ii) playing around with conditions so that this system peak, water dip or whatever you want to call the think comes at a significantly later retention time.

Thanks very much,

Dave

I'm no expert on TOC detection, but here goes...
As far as reducing bleed, I suspect that replacing as much of the tubing as possible w/ stainless would be a good start. Then there are the seals. Different vendors have different grades of seals available for pumps (generally), so it could be worth investigating different seal compositions (Optimize, for example).

That said, reducing bleed from the LC may be like removing deck chairs from the Queen Mary to lighten the load. I'd spend time looking for columns with more stable stationary phase...

Sounds like a problem with no perfect solution. One idea is to place a short column between the pump and the injector. There is a risk that this will contribute to the carbon background but I'm thinking it may sorb most of the "junk" that bleeds off of the mobile phase reservoir lines (mostly phthalates I think) and thereby do more good than harm. But my strongest inclination would be to simply run the system for a few hours before each time that you plan to use it, in order to allow the carbon background to reduce and stabalize. During the last half hour of this interval you could collect the effluent and then use that to prepare your sample.

Hope this helps.

I'm wondering from the description if your problem might not be related to dissolved CO2 in your mobile phase (with less being present in you're sample, hence the negative peak). Try injecting sodium bicarbonate and see if you see a peak at that retention time. If so, you could trap the CO2 in your eluent with a anion exchange trap in the OH- form or change retention with a pH shift of your eluent. What is your eluent in this system?

Thank you for the time you have taken to help me with this. I actually had the resident Notre Dame mass spec specialist see if he could identify anything in the HPLC stream and indeed he said he was very sure he was seeing pthalate. I think swapping as many of the lines as poss. to stainless steel tubing is a good idea. Putting a column between the pump and injector is also an excellent suggestion.

I promise to report back with a post on this forum when we have tried these approaches.

David