Verrry late baseline disruption/elutor(?)

[Robotjock]

Good afternoon All;
I have a very interesting problem. I am developing an assay/impurity method w/ following parameters.

column - chirex 3125 d-penicillamine 250x4.6 5u
MP - 1mM Cu(II) Acetate (aq)
flow - 1 ml/min
Col temp - 25C thermostatted
inj vol - 20 uL of 2 mg/mL sample
Detection - 230 nm, VWD
Run time - 40 minutes
Sample: N-substituted L-arginine
HPLC - Agilent 1100

Injection of the 2 mg/mL sample leads to a baseline disruption 7 hrs later.

I have confirmed this w/ duplicate experimental studies where five samples were injected using above conditions, then the LC system switched to a 480 minute run time method. In each case, the baseline drops ~ -10 mAU 7 hours laters, then recovers over 40 minutes corresponding to the assay method run time. The initial drop is not sharp, but steep enough to make integration of trace peaks difficult.
Comments

Thanks,
Robotjock

[tom jupille]

In a sense, not surprising. In principle, there will be one negative peak for every component in the mobile phase if a sample which does not contain that component is injected. I learned this the hard way many years ago when I was doing "single-column" ion chromatography (direct conductivity detection with no suppression; the company that I worked for manufactured conductivity detectors). We had a customer who complained that he was getting negative peaks for low levels of chloride. On further probing, he admitted that what he was seeing was a calibration line that had a significant negative zero intercept, so that if he injected a DI water blank, he in fact got a big negative peak at exactly the right retention time for chloride.

It took us a couple of weeks of playing around before the light finally dawned: when preparing his mobile phase, he had adjusted the pH by putting the electrode directly into the mobile phase buffer. His electrode had a leaky junction, and he was ending up with some KCL leaking into the mobile phase (about 1 ppm, as I recall). When he injected DI water, that 1 ppm deficit showed up as a "vacancy" peak for chloride.

At the time, we were using Santa Clara tap water as a quick and dirty test sample (it had measurable amounts of chloride, nitrate, and sulfate) to test retention times. On a dare, we made up a batch of mobile phase buffer using tap water and injected DI water. Sure enough, we got a an "inverse" chromatogram; all the peaks at the right place, except negative rather than positive.

I've maintained for years that the reason we aren't plagued by negative peaks is two-fold:
- our detectors are blind to most of the mobile phase components
- on sheer statistical grounds, most of these "system peaks" are either unretained (in which case they simple show up as part of the t0 baseline upset) or so strongly retained that come out hours later as broad peaks which we don't notice.

If I had to guess, I'd say that what you're seeing is the system peak from cuprate, and I'd be willing to bet the price of a pizza (cheapskate that I am) that it would go away if you added the Cu acetate to the sample.

[Kostas Petritis]

I completly agree with Tom, except maybe from the last part as even if you add the Cu acetate to the sample you will still maybe see such a peak after 7 hours due to the distrurbance of the equilibrium because of your analytes (if this is true, I will claim my pizza next time I'll see you in a conference or so ).

I have similar experience as Tom has but mostly when working with ion-pairing chromatography and conductivity detectors.

Have a look in the following article: K. Petritis et al. J. Sep. Sci. 2001, 24, 397-405 (Ion-pair reversed phase liquid chromatography-indirect conductivity detection and response deviations of underivatized amino acids).

In Figure 2, you will see exactly what you are describing. At that figure I prove that the negative system peak is corresponding to the equilibration time of the column and you can actually see that the peak becomes more and more wide the longer the column equilibration. After 7 hours I would expect something barely noticable as what you are describing.

In the same article you can also see that the peaks can be positive or negative depending on their charge and depending on if they are eluted before or after the system peak. Furthermore, I use these properties in order to quantify a fermentation product that contains 1000 times more glutamic acid than other amino acids. By operating very closely to the isoelectric point of glutamic acid I almost complety quence it's conductivity response allowing the detection of nearly eluting polar amino acids.

Anyway, the article if full of useful discussions and relevant references that covers a lot of what Tom just mentioned.

Hope the above helps!

Kostas

[Robotjock]

Tom & Kostas;

I appreciate the replies, they are very helpful.

The thought has occured that the conditions might be "right" for indirect detection/vacancy LC. However, my samples are diluted w/ the MP and, thus, contain Cu (II) Acetate. So something else is the cause. I'd take you up, Tom, on the price of a pizza, but I should have stated the sample diluent initially.

Kostas - your column equilibration cause is an interesting theory. The column manufacturer does state a 2 hour equilibration time is reccomended. I have some questions but will reserve them until I read the article.

Thanks again,
Robotjock

[tom jupille]

That's why I limit my bets to pizza money!

If either (or both) of you guys are going to Pittsburgh
Conference, e-mail me and I'll arrange to pay up.

[Robotjock]

Kostas,

I have read the article. I am not sure I can show definitively that the baseline disruption is a system peak. I don't have the luxury of changing the mobile phase additive. Well I can use Cu (II) w/ acetate, chloride or sulfate, but as the Cu(II) is the critical component I don't believe the anion will make much difference. May be by varing the concentration of the Cu (II) cation, 1 mM, 3 mM, 5 mM etc., I'll see decreasing equilibrium times.

Even if this would show/prove that the baseline disruption is related to column equilibration, still have to determine a work around to permit accurate peak integration.
regards,

[Kostas Petritis]

Robotjock,

There are ways to show that this is due to the system relaxation but it seems that this is not your main concern.

If sensitivity is not a problem (although I guess it is as you talked about trace amounts) I would advice you to inject smaller concentrations of your sample. That will limit the base line drop which might be enough for accurate integration.

There is another solution... you can calculate your analysis time in order to have your baseline disruption to occur at the end of your analysis (in the detriment of your throughput). For example if the disruption is after 400 minutes and is occuring for 10 minutes you could increase your analysis time by adding a post-time injection of 20 min. Now your first 6 inejctions won't have any disruption (but you still have to let the 20 min post time). The seventh injection (starting at 360 min) will have the disruption after all the peaks of interest are eluted (i.e. 40 min) and it will not interfere with your analysis. From their after you will always have the disruption after 40 min but it shouldn't intefere. (The catch is that you should consider these disruptions as chromatographic peaks... a peak that is eluted after 7 hours it's retention time might vary significally as after 420 min with a +/- 1% retention time variability makes +/- 4.2 min...).

Varying the Cu concentration might give you or not different equilibration times (it depends).

[Markus Laeubli, Metrohm]

Robotjock,

As I also work since longtime on non-suppressed ion chromatography, I am pretty familiar with system peaks. The most easy way to check them is to inject the mobile phase and as a second injection ultrapure water. The first injection should minimize the system peak, the second should give a high (neg. or pos.) peak.
In IC the system peak is usually depending of the total concentration of the ions. In very early times of non-suppressed IC the system peak therefore was interpreted as carbonate.
The bad news about a system peak is that there is no way around it.