I would appreciate any insights on a problem I've been having with a particular compound. I'm currently using a Discovery C8 column and mobile phase consisting of 80% 20mM Phosphate buffer (pH 7.8), 10% Methanol, 10% Acetonitrile. My peak of interest fronts badly at low concentrations and recoveries are low at these levels. y-intercepts are consistantly below zero. Higher concentrations look great and show plates around 10,000. In development work, fronting seemed to improve when higher pH was used but never went away completely. Other peaks appear to be ideal (no fronting, etc.). Literature regarding this compound suggests that it will dissociate in water to form degradants which happen to elute earlier in the chromatogram. All I can think of is that this is happening on column to result in the fronting. Normal phase is not a viable option because of the nature of my sample and I'd also like to avoid derivitization. Any advice? The compound is a mercury-containing preservative. Can the mercury be doing something on column active sites?

You may want to look at increasing the ionic strength of your buffer to see if this has any effect. This may or may not help.

How "bad" are your recoveries at low levels, and how "good" are your recoveries at higher levels, percentage wise? Are the recoveries bad enough that you would expect to see a significant difference at the higher levels?

Just some thoughts. Good Luck.

Hi! I knew I'd hear from you. I did try increasing the buffer concentration fivefold and, interestingly, the fronting got worse (retention time unaffected). Eliminating the buffer altogether made peak elute much earlier, which wasn't really a surprise, but there was still some fronting (not as much) and selectivity lost. Reducing buffer concentration to 10mM didn't seem to help either.
Recoveries were excellent at the higher levels (from about 10 ug/mL to 50 ug/mL) but below that the peak shape seems to undermine the area response. Even at higher concentrations, a slight "foot" is observable near the baseline. As concentration decreases, the peak distortion is increased. This does not just occur in degraded samples, as I read about in literature, but also in diluted standards.

You seem to be associating the low "recovery" at low concentration to difficulties with the integration rather than a true loss of compound. Is this correct?
If indeed the analyte decomposes in the mobile phase, you should be able to verify this by dissolving the sample in mobile phase and doing a kinetic.
It is possible that the analyte can convert reversibly to a compound with a related structure. Anomer separations of sugars or structural isomers around a proline in peptides are examples of this.
Or maybe there are some simpler explanations....

I actually have associated the low recovery with both difficulties in integration and true loss of compound. On the first point, the peak shape badly deteriorates until it is indistinguishable from the baseline. On the latter possibility, I thought that perhaps only a finite amount of degradation is occurring during the elution time and that lower concentrations are therefore more affected because the amount of degradation is proportionately increased.
As to your question about the reversible conversion, that is supposed to be what is happening according to literature. My compound supposedly dissociates reversibly but the same article said that spiking the sample solution with degradant did not drive the equilibrium toward the analyte (LC results were not shown).

It sounds like it is a peak shape issue. You have seen some fronting and this "foot" at higher concentrations as well. If you haven't already I would look into your dissolving solvent and injection volume. If your sample is in a predominantly organic solution and you have injection volumes above 10uL, fronting may be the result since you have a high aqueous mobile phase.

I would also try increasing the column temp and see if you have any peak shape improvements.

Also, if you are detecting with UV, try using a PDA and checking for peak purity.

Just some suggestions, hope it goes well
Daren

can it be that your decompostion-produkt is co-eluteing with your analyte, and desturbing the peaksheap that looks like a fronting peak? with a PDA or MS-detector you can look for peak-impurity as daren said...
if the respons of the decomposition-product is lower then the respons of your analyte then this could also explain the bad recovery's

Thanks for the input, Daren and Anonymous. Unfortunately, I don't have access to a PDA. Youre suggestions are good, but my sample is in water, which I failed to mention, so that probably means I don't have a solvent compatibility issue. However, your injection volume idea (Daren) has made me wonder what would happen if I made a smaller injection of a more concentrated solution.
As to the possibility of a coeluting decomp product, it would have to be because more decomposition occurs at lower concentrations, and I'm sure that is a possibility. But changing organic content, column type, pH, and temperature has not helped much.

If I understand the background information correctly, the analyte of interest ( we will call it 1.) converts to something (which we will call 2) And I think you said the conversion of 1 to 2 is reversible under the conditions of chromatography, that is to say they are in equilibrium when present in a solution. And, 2 elutes before 1.

What you will see in the chromatogram will depend on how rapidly 1 goes to 2 and 2 to 1. If the rate is not very much faster than the time it takes to do the separation, then 1 will never achieve equibrium with 2 during the separation and therefore, all during the separation 1 will be converting to 2, because 2 is continuously separatied from 2 by the column. The result will be a fronting peak and there is nothing you can do about it unless you do something to alter the kinetics. If 2 has a different response factor than 1, quantitation is doomed, because you can't simply sum the peaks and you also can't ever separate them in the integration.

I have a fair amount of experience with these situations and you may email me for further information if you wish bjtindall@juno.com
In any case, the solution is to understand the kinetics of the reaction and alter these kinetics by some means-pH, temperature,??- to get chromatography you can use.

what about trying to make the seperation better, determin the rel.respons factor from compound 2 to compound 1. the you can add the reponses: (rel.repons comp.1 to compound 2) *repons compound2 + repons compound 1.?

Anonymous,
If 1 is converting to 2 during the separation with kinetics similar to the time scale of the separation, you can never separate 1 from 2. As 1 travels down the column it is generating 2 all during the separation right up to the end of the column (and beyond). So, the last 2 generated before 1 elutes is still under the peak for 1. Equilibrium is never reached because the separation removes the product from the zone of 1.

An example of such a situation occurs in the separation of anhydrides, for example acetic anhydride(see J. Chromatogr. 868, 41(1999) Fig 4). Without more explanation of the chemistry occuring during the separation in question I don't know if the above example is relevant of not, but if it is the only solution is to do something about the kinetics to make them faster or slower than the separation. Or pick another method.

As far as I know, sample overload, or sample decomposition to a less retained compound, are the only causes of fronting peaks.

Bill,
This is not so simple, and it depends on the details of the kinetics.
If the parent analyte reacts into a daughter analyte on the column without a reversal, the front of the peak will have a different spectrum than the parent peak, and the spectrum of the parent peak will change over the peak. If we are slowing down the c-graphy by changing the flow rate, we will get more of the daughter compound and less of the parent.
If it is a reversible reaction, the question is at what point in time the reaction starts, after the injection or already before the injection in the vial. If we have equilibrium in the vial, we have the same situation as the classical sugar anomer separation. Slower flow rates will make the problem go away and one gets a more symmetrical peak, while very rapid separations may ultimately result in two peaks, which one may be able to distiguish via the spectra. If it is a reversible reaction that starts only after the injection (for example if the sample was dissolved in something else then mobile phase), the situation is very complex and depends on the details of the kinetics.
What bother me about this is the concentration dependence reported by Bill Scheidt. The problem may have nothing to do with kinetics...

Could there be something in your preparation that is reacting with the analyte? If there were something present at sufficiently low concentration, it could be reacting until it is consumed. At higher analyte concentrations the amount of this reactant would seem less due to the greater size of the parent peak. At lower concentrations, you would still form the same amount of the daughter peak because the impurity concentration would control the extent of the reaction but the amount of the daughter peak relative to the parent peak would be greater and the fronting would appear to increase. If this were the case and the reaction proceeds to completion (no equilibrium process occurs), there could be a dilution at which the daughter peak becomes greater than the parent, though this may be too low to be detected with your method.

Thank you all for the valuable and very insightfull input. This week I tried a couple things without much success: 1) I experimented with mobile phase pH... phosphate buffer at pH 6.2, 7.2, and 8. Fronting improves somewhat from 6.2 to 7.2, but increasing to pH 8 made less of a difference. My current column cannot go past pH of 8, so that's where I stopped. 2) I tried using a subambient column temperature of 15 degC. This increased the RT and the peak width, and fronting did not improve. I had hoped that one of these two changes would effect the kinetics. One thing I haven't tried (but will) is something suggested by Uwe, which is to experiment with flow rate. Anonymous, I have had the same thoughts that you just brought up, but not sure how to design experiment for that. My standard solutions are prepared in distilled water and I haven't tried anything else for a diluent because samples (vaccines) are aqueous and because the mobile phase is 80% aqueous. It seems to me that any reaction/equilibrium must be taking place prior to injection but perhaps there is also an equilibrium shift once it hits the column. Thanks again for info/help.

One can imagine another possibility to explain the concentration effect: your column has active sites which shift the equilibrium toward the early eluter. The sites are, of course limited so that at lower concentration the proportionality is higher. Have you tried other columns?
When one has a problem like this a lot can be learned if peak fragments are collected and reinjected. . . . .
If you do this, please tell us what the results are.
See also the chain "Fluorouracil and Buffer vs Water for Mobile Phase" Jan. 24.

That sounds like a great idea. I don't have a preparative system; do you think I can get enough collected on my analytical system? I'm using an HP1100.
I have also considered the possibility of active sites. Something that I recently discovered seems to lend evidence to that theory. When I pump mobile phase for a while and then make standard injections, the first injection always has a lower area response than the next. I tried adding TEA to the mobile phase, thinking that I could block active sites that way, but I saw no improvement.

Since you see the peak you have enough material. You may have to concentrate the fractions a little. Loosing more organics than water could be an advantage. But, if you can, use a rotary evaporator, not N2 (have seen huge gas peaks with N2 . . .).