Hello,

We are 2 danish lab-techs with a huge systempeak problem !

In developing a new analitical method there apeard a systempeak with unknown origin.
We use HP1100 HPLC systems and we have a mobile fase witch i mixed over a binary pump. Our mobile fase contains 55% MeOH and 45% 30mM Phosphatebuffer pH=7.5, and we use a Waters Xterra column 250*4,6mm with 5µm partikels
We see the systempeak in both blank(water and mobile fase) injektions and samples. The peak inceases with higher injectionvolume.
We tried to use other HP1100 HPLC's in our department but we still see this peak. We investigated if it could be the MeOH and water used for the analysis, but it was not. We also investigated our vails, caps, columns,transferpipets,autosampler (by using manual injection) pumplubricant and our glass.

We can't figure it out anymore and are becoming a little desperate. Is there someone out there who can help us ?

Have you tried changing the solvent in the needle dip vial. It can (and will) become contaminated over time and can cause problems like you are seeing as well as carryover issues.

What is the RT and peak shape of that peak?Whether it is broad or sharp?Have you tried equivaqlent column in the same mobile phase?If u increase the flow rate whether there is any change in peak shape or RT?

We are sure that it is not contamination from the needle dip vial. The retentiontime is 6.7 minutes and the dead volume is app. 2.5 minutes. The peak is nearly a perfect peak that is very stable. If we inject 5 times, the systempeak is very much the same (%RSD down to 0,07 %!) We have tried using a equivalent Luna column, and the peak is seen there as well. Since yesterday we have also tried running the system on a Shimadzu HPLC, and we have the same systempeak there.

The peak change Rt with changed eluent, so it must be a "real" peak.

We're now suspecting our Methanol. If I inject pure MeOH the systempeak gets 6 times bigger. Have anybody outthere experienced any difficulties using Fisher MeOH. We have ordered some other brand, but we haven't got it yet.

Can anyone explain why we see a peak, when injecting water or mobile phase.

(If anybody wonders, who I am: I'm the other danish technician.)

Did you inject degased MF? It could be oxigen dissolved in undegased water, methanol or MF.
If you search your past isues of LC-GC magazine you'll find more about this in one of LC-troubleshooting column.

If you are not running a gradient, try premixing the mobile phase and running with just one pump.

If you are using gradient elution, try running the starting conditions for an extra 10 or 15 minutes before injecting a blank. If the system peak is larger after the longer time at starting conditions than it was/is after normal consecutive injections, you have a water problem. Either filtering your aqueuos phase through a C-18 sep pak before use, or putting a cartridge guard column between the aqueous phase pump and the mixing chamber should help (this too has been discussed in LC*GC).

If you are running a low pressure mixing gradient, degas mobile phases thoroughly while shopping for a high pressure mixing solution.

I agree with Andrej.B.
The easiest way to make an evidence, that different oxigene concentrations in mobile phase and sample(blank)causing this system peak, is just to collect mobile phase from the detector outlet and inject it again.

Mr. DR's final comment has me intrigued, is he implying that running gradients on low pressure mixing systems does not work or that these systems contribute extraneous peaks. Maybe the message is that high pressure mixing systems offer better performace in all cases?

Anyone else like to comment?

Just wondering.....

Dear DR,

HP1100's are high pressure systems, see message #1

Low pressure mixing systems do work for gradients. The concern in gradient runs using low pressure mixing is the same as with isocratic runs using low pressure mixing to make up your mobile phase composition. Both solvents must be thoroughly degassed using low pressure mixing. This is because mixing of different air saturated solvents, i.e. ACN/water, will cause degassing and air bubbles in your mixing chamber, pump heads, check valves, etc. This can cause in a sudden drop in pressure and no flow. As far as reproducible extraneous peaks, this is not a result of low vs. high pressure mixing. I have seen system peaks during gradient runs, but these were not proportional to injection volume. They were a result of system contamination. Over all, an opinion only, high pressure systems are more forgiving.

It really sounds like a contaminant slipped in somewhere, somehow. I would suggest starting from scratch with all new solvents (different bottles), buffers, CLEAN glassware, etc.

I might add to Anonymous' explanation that any gas is LESS soluble in a mixture of liquids than it is in either of them separately. Therefore, two mobile phase components individually may be less than saturated with atmospheric gases. When combined, however, they may be saturated or supersaturated. When exposed to negative pressures as in a low-pressure mixing situation, such a saturated solution will probably outgas with all the attendant problems.

I must say that we're very happy that you all give good suggestions, but:

We have tried injecting degassed mobile phase and the same systempeak appears with the same area as normal. We see the systempeak in variable areas in both:
- Mobile phase
- Methanol (greatest area)
- Acetonitrile
- Milli-Q water
- Hplc-water from an external company
- Using our normal vials
- Using another kind of vials
- Using different brands of columns
- Using both Shimadzu and (3 different)HP-1100
- Using new bottles of solvent
- Using mobile phase mixed by pump AND premixed by us
- And (ofcourse!!!) tried from scratch several times

I still think it is the brand (Fisher) of methanol, but then I wonder why we don't see it in our other methods?!? It is NOT a new problem in this method, so it isn't just a batch of methanol from Fisher going wrong...It's several batches.

Please...Give us more suggestions.

Re: Low pressure gradient formation - I'm not saying that it can't be done - I do agree that it tends to be much more forgiving.

What you have done so far does not rule out a water problem. Bottled LC grade water can cause peaks as can Milli-Q water when a few styrene bits shed by a DI cartridge get through our best technological road blocks (0.22µ filtration after a UV treatment). At low UV wavelengths they can be seen. Mobile phases with low organic concentrations allow them to build up on the column head. I maintain that trying a bit of in-line C-18 filtration, particularly of the more aqueous portion of the phase be tried. Hopefully, it will work - or at least minimize the problem...

Regards,
DR

Did you consider the phosphate buffer?

I had once a problem like You and we made a thorough investigation and couldn't find anything first. It looked like Your problem the peak was there all the time. Then we analyzed each step and found one step that were the same for all the injected solutions. The transfer of the sample to the HPLC-vial. We used a special brand of disposable glass-pipettes. To test the idea we filled the vial directly from the waste-outlet from the system and the peak disappeared. When we rinsed the pipette with the mobile phase the peak reappeared. We began to use another type of disposable pipettes and the problem disappeared. You say tested your transfer pipettes, how did you test them?

We tested our transferpipettes by using two different types, glass and plastic. Here we didn't see any change whatsoever. So we desided to fill the vails without tranferpipettes.
The peak was still there, just as it always had.
This rulled out the transferpipettes.

Yes, we considered the phosphatebuffer. But we still can't explain why the peak appears by just injecting water or MeOH !!!!

Please keep coming with advise, we're rulling out posibilities here, so we might come closer.
Thanks !

Sometimes 0.45µ filters used for filtering the samples or buffers may cause this problem.

Have you tried a blank run, i.e. start the cycle without actually making any injection at all?

You can do this on an HP-1100 by using the single sample option (as opposed to sequence option) for injection. Let the vial number be blank and you get no injection at all, but the cycle starts as usual.

This test will show whether it's the injection of a sample causing the peak or something in the instrument itself. From what you say, I'd expect to see no peak in this test.

I remember that one batch of new HP vials were contaminated. Are you using HP vials? If so, try first to clean vials with Methanol and ultrasonic bath.
Oxygen could also be the problem. Try degassing sample with He or N2.
Do you have DAD detector? Can you take UV on-line spectra?

Again: Thanks for the advice, but we're still not there. It's not the Phosphatebuffer, because we've tried to use ammoniumacetate instead. There is still a peak. It is not the vials, because we've tried three different kinds of vials.

If we inject nothing we do not get a peak, so it's not the system. If you see above, we have already tried injecting manually without using the autosampler at all.

Is it the water? I'm not sure because we get the biggest peak injecting methanol. If we inject water we only get a peak with a area of 17% of the methanol-peak-area.

We have now tried another brand of methanol, but we still get the peak. If it's oxygen, i wouldn't suspect it to be so repeatable as this is, We have RSD down to .07%!!!

The UV-specter almost looks like baseline, there's just a bit more in the low end (200-220 nm)

We cannot ionise it on our MS, mot with Electrospray, neither APCI.

I've gone back through your postings and I don't find that you've said what wavelength you normally monitor. In response to Marcelo, you seem to be saying that the peak shows absorbance only from 200-220 nm. Is that where you're monitoring? If so, then my next question is what is the lambda-max of your analytes?

If you have been montioring at less than 220 nm, you could be seeing the methanol you're injectiong -- not a contaminent, but the methanol itself. However, I don't think this would fully explain the symptoms you've described. For example, injecting pure water should then give you a negative peak! Furthermore, I doubt methanol would be retained appreciably in your system. So a contaminent seems more likely.

However, such low wavenlengths will reveal all sorts of things we chromatographers would rather keep hidden. Most of the time, you can choose a wavelength to show what you want to see and hide what you don't, and your chromatography is none the worse for it. (Large or very active hidden components can give you nasty surprises, however.)

If it is a water problem, it makes sense that a higher % organic slug (ie a methanol injection) might reveal the system peak. If it is something in the water, it will stay on the column until the % organic is sufficient to push it off. This is why I suggested running a few blank injections and comparing those system peaks with those from injections using a longer re-equilibration time. If the water is the source, you should see larger system peaks after the longer equilibration times.

This all assumes that you are running a gradient.

We detect at 225 nm, but we have a DAD-detector, so that's why we know what's happening in the low end.

In respond to DR, I can say that we have tried running blank injections, and there's no change. We do see the systempeak, both when we're running a gradient, AND when we run isocratic. The Systempeak appears at approx. 7 min, and the gradient first starts at 10 min, so we've run a lot of samples only running for 10 min, that is wihtout the gradient!!!

You inject nothing ("blind injection") - no system peak.
You inject ANYthing either manually or automatically - regular, reproducible system peak.

Aren't you working in an "unhealthy atmosphere" of some organic volatile compounds which may be absorbed from the air, more likely to an organic solvent (like MeOH) rather than water? Did you inject fresh MeOH and water DIRECTLY from the newly opened bottles?

Sorry if it may seem funny, but this situation looks quite interesting.

Please look at the testchromatograms for the Xterra Column!!! There is a systempeak in Waters own testchromatograms, precisely where we see it. It looks like the new Xterra-column is having a bit of a problem. The systempeak-mystery is hereby declared solved!

In a previous message you mentioned that you had tried a run(s) using a Luna column and still saw the system peak. Perhaps the mystery is not totally solved yet...

A.A.

Assuming that you wouldn´t rouse up the whole chromatographing world by forgetting that in UV detection one usually gets a peak due to the shock front after an injection (whose position does also somewhat depend on mobile phase), I can only think of a common problem in our lab: Rheodynes collect some substances between the stator and rotor (ie, cortisol. ouabain...) which are only very slowly relased again during many subsequent injections. You can only get rid of this in reasonable time by taking the Rheodyne apart and thoroughly cleaning the rotor and stator.
Also I am assuming that you ruled out carryover in any other part of your system.
To all participants: if you have a riddle of larger dimensions. please give sufficiently precise details the first time around.

I have been pursueing the same problem you've experienced and I also was beginning to look into dissolved oxygen, before I decided to try this forum. This may be a combination of two problems. One is the dissolved oxygen in the samples and secondly, I was told by a HP representative that the autosampler is designed to pull an additional mm after sampling the vial, leaving a small air space at the tip of the needle. This is purposely done so that if a needle wash vial were to be used that you wouldn't contaminate it with sample at the tip of the needle, or so I'm told. I wasn't told how to fix it though. Supposedly an injector program will work but I haven't had any success doing so. If this is true, taking the autosampler out and injecting manually with a Rheodyne valve should work.

By now you have probably solved the problem. If not, what about using freshly prepared phosphate buffer. Buffers in general can cause bacterial growth and these bacteria release compounds into the solvent which is then observed by the HPLC's detector. - Just an idea.

To add weight to the "Dissolved oxygen" theory, I too had a similar problem - anything I injected had a system peak which behaved as a real peak (increased with injection volume etc). The peak size was greater with organic sample solvents than with aqueous, and was reproducible from one injection to the next. I know you have tried degassing your sample solvent, but how you do it and how long you leave it before injecting is the key. In order to get rid of the peak, I had to degass the sample solvent by refluxing for at least 15 minutes (helium sparging proved ineffective) and injecting immediately. By degassing even more (refluxing for about 40 minutes) and injecting immediately, I could actually induce a negative peak where the system peak usually was. However, if you leave the sample solvent sitting even in an autosampler vial for 5 minutes, it seems to re-absorb oxygen immediately to the same saturation level as before. By covering the surface of the sample solvent in the vial with a layer of immiscible solvent, I could stop the re-absorption on oxygen for a reasonable amount of time, but this was really no practical solution. In the end I had to learn to live with it and re-develop my method conditions to stop it interfering with peaks of interest.

Your problem seems so similar to mine - do let us all know if you find a way to get rid of this peak for ever!