"phantom" peak

[Jackus]

Hello everybody.
I found plenty of very clever men here so I hope you can help me with problem I have been faced with.
We test purity of a very sensitive substance. Unfortunately I cannot send structure of the product.

Chromatographic conditions are:
Column: Silica based C-18, 250x4.6 mm
Mobile phase: Acetonitrile/borate buffer, pH approx. 10
Phase used for dilution: Acetonitrile/20mM solution of glycine (added for better stability of product)

Substance is very sensitive and unstable in neutral or acidic conditions. As stabilizing agent NaOH is added into product (crystalline powder).
Sample is weighted, diluted in "dilution phase" and injected.
Solvent peak elute at 2 min.
Main product elute at 12 min.
Some batches contain impurity which can be seen at 2.7 min. (my "phantom" peak) and it is source of my headaches.
The "phantom" peak has been found only in several batches. When this batch is re-tested its amount is the same. If the same sample is injected after 15 minutes the peak disappear.
Till now you can say OK, you have unstable contaminant, but...

- UV spectrum is the same as main compound (99% match - Chemstation)
- It seems that product of decomposition of my "phantom" peak is the same as product of decomposition of main compound
- When I did semipreparative injection, quickly re-analysed collected "phantom" peak I found only two peaks: product of decomposition and ...
... main compound!

My result is that small amount (up to 0,1%) of main compound is teared-off during injection and elute practically unretained. But why is it observed only in some batches and on the same level?

Have you seen such strange chromatographic behaviour? Do you have any other explanation for me? So far I have thought that I know quite a lot about chromatograpy but somebody on high kicked my self-confidence ...

Thank you for comments and forbearance to my english.

[danko]

Hi Jackus,

Several thoughts from here – mostly pH related.

First, beware of rapid stationary phase support degradation/dissolution hence mobile phase pH = 10.

Second, you didn’t tell what the sample solvent’s pH is and which injection volume you’re applying.

Third, I can only speak for my self, but I think we need the ACN : water (the aqueous part) ratios in the mobile phase, as well as in the sample solvent.

Here is what I’m thinking of – currently: The mobile phase (especially injected in a large volume) could be much stronger either due to its pH or ACN content, which could cause a rapid elution of a small part of your analyte, prior to the re-establishment of the initial conditions in the column. Why seeing this with some bathces only? Well, it might be a synergy between sample matrix and sample solvent which is not always completely alike?

Finally a comment about UV spectra: While it’s a quick and easy test of compounds’ similarity the method is very unspecific – especially when dealing with related compounds, like metabolites of a substance.

That’s all for now. But if you share some more parameters with us (e.g. diverse pH’s, mobile phase’s and solvent’s ACN content, flow rate etc.) it can turn out to be something else.

Best Regards

[danko]

Correction:

The mobile phase (especially injected in a large volume) could be much stronger either due to its pH or ACN content........

I meant of cours the sample solvent - rather than the mobile phase!

Best Regards

[Jackus]

Hi danko,
thank you for your reply. I'll try to answer your questions.

First, beware of rapid stationary phase support degradation/dissolution hence mobile phase pH = 10.

We use Zorbax Extend C18 column which is stable at pH 9-12. I can confirm good stability and long life of that column.

Second, you didn’t tell what the sample solvent’s pH is and which injection volume you’re applying.

Both solvents are nearly the same pH about pH10. There may be small difference because pH is adjusted in aqueous solution, then organic component is added. Injection volume is 10ul, concentration 1mg/ml.

Third, I can only speak for my self, but I think we need the ACN : water (the aqueous part) ratios in the mobile phase, as well as in the sample solvent.

Both solvents are set to have approximately the same content of organic compound. It is 20% of acetonitrile.

I'm not so sure about UV spectra similarity too but it is only one possibility I have. I think that even LC-MS is not sufficient tool how to obtain more information due to buffer incompatibility and low amount of unknown compound.

Thank you for your time.

Best Regards

[danko]

OK – more complicated than I thought.

I have a suggestion for you: Dilute you normal sample solution (from a batch that shows the abnormality in question) in the sample solvent, so that you have one solution containing half as much of the analyte as the normal solution.
Then inject the normal solution twice or three times and the diluted solution (twice the normal volume i.e. 20 μL) as many times.
The test will show (reproducibly) whether or not you’re dealing with a solvent related issue.
Maybe you’d like to get back and tell what you observed.

Best Regards

[Jackus]

Hello danko,

the problem is even more complicated then you think now.

As I mentioned earlier the impurity is unstable and disappear within 15 mins after dissolving. I'm not able to make more injections from one solution than one. When I prepare several weighings of the substance and each is diluted and immediatelly injected then amount (area%) is constant no matter what volume is injected (and what is concentration of the solution).
I know that behaviour of "my" impurity is strange. It is a reason why we officially call it Phantom peak in our documentation. We did several studies and experiments trying to find whether it is remainder from previous reaction steps, we tried different batches of used chemicals without significant influence to impurity content.
One crazy theory is shifted to another...

Edit:
Chromatograms tells more:
Semipreparation. When collected fraction (3) was injected to anther instrument only peak of main product was found. Following fraction is without it so it cannot be tailing from previous injection.


Here you can see rapid decomposition of phantom peak. The second injection was performed within 20 mins after dissolution.

[danko]

Hi Jackus,

It's fun - I must admit.

Would you please, following a sample elution, inject a couple of blanks (i.e. sample solvent) and a couple of zero microliters injections as well - under the same conditions as above?

Best Regards

[Jackus]

Hello danko,
thank you for patience. I did such test earlier to find whether it is not some effect related with pressure drop or phase mixing during injection. Phantom peak was not found.

[danko]

No problem. Patience is a good property in trouble-shooting situationes

I thought you could've done the test I suggested, but I needed to be sure.

What about fraction 2 - did you re-inject that? Did you perform an UV scan on that?
Without the possibility to zoom in on the chrom., the only change I see (appart from the decreased area of the "phantom peak") is a peak-like signal in fraction 2.

Best Regards

[Jackus]

Here is my fraction 2.

You can see traces of main product at 14.02 and it's main decomposition product at 3.32 min.

I found UV spectra of both compounds (main compound and my "phantom").

(I'm deeply sorry for quality of the picture. I copied it from documentation).
The main decomposition product has absolutely different spectra (match is about 60%).

[danko]

I assume the last chromatogram, shown above, represents fraction 2 of the decomposed compound. But what happens if you re-inject fraction 2 while the compound is intact (i.e. the first injection 07-May-10, 15:19:58)? Do you see the same pattern, or haven’t you tried that?

Another suggestion: You write; the substance is very unstable, especially at neutral and low pH. So, I thought; what would happen if you exaggerated the sample environment in order to facilitate more degradation (e.g. one solution at pH 10 as a reference, then one at neutral pH and finally one at acidic pH. Ideally, the degradation rate will follow the pH decrease and the peaks representing actual degradents will rise.
So, what do you think?

Best Regards

[Bruce Hamilton]

What happens when you analyse a sample of batches with the phantom peak when the sample is injected immediately after preparation only in mobile phase, with half/twice the concentration of glycine, or even in water.?.

Borate forms complexes with a wide range of molecules, I'm wondering if trace formulation differences are interacting with the mobile phase buffer.

Bruce Hamilton

[Jackus]

Hello guys. I'm sorry for my delay.
Danko, shown chromatogram represents fraction 2 which was collected during first injection but as I wrote earlier the unknown inpurity decomposes very quickly.
The substance is really unstable and when we were developing the analytical method we found that decomposes within minutes even at neutral and slightly basic solution. High pH above 10 and presence of glycine guarantee stability of solution at least 3 hours at 4 °C.

During degradation of the impurity there is not any other peak increasing except degradation product of main compound (i.e. there is not any other degradant which may be from decomposed impurity). Is "degradant" real english word or I created a new one?

Bruce, when I injected the sample containing phantom peak diluted in 20% ACN/water (without glycine, pH neutral) the phantom peak is still present but at lower concentration (about 2/3 of original). The amount of decompositon product is higher several times.
Borate buffers are really strange and I hate them not only because of forming complexes but amount of repairs I have to solve (valves are leaky and needs replacing once per 1-2 months even if mobile phase is filtered before use ). Now I cannot remember but I think I did some tests with different buffers. I'll search for a chromatograms tomorrow. Please be patient.

Have a nice day.
Regards

[danko]

Danko, shown chromatogram represents fraction 2 which was collected during first injection but as I wrote earlier the unknown inpurity decomposes very quickly.

That makes it even more interesting to see a chromatogram of fraction 2 collected during the second injection and the third (if possible) and so on. Because that might indicate what happens with the peak with retention time 4.916 over time.

I still think a forced degradation at neutral and low pH followed by a fractionation could generate a valuable info. helping you understand the nature of the “phantomâ€