Chromatography Forum

Scenario:

A lab with a long standing tradition of utilizing external standards (in water) for analysis of sugars (glucose, fructose, sucrose, maltose) in various fruit juice concentrates - a uniform method has been developed for this that does display adequate resolution for these compounds in the external standards. Detector is RI

After being questioned on a confusing result it has been discovered that (ah ha) the various concentrates have additional peaks that often do not have appropriate resolution from the peaks of interest - < 1.5

Leaving aside the possibility of actulal co-eluters for the moment, I actually see our only options as
1 - trying to develop a separate method for each product (literally dozens) and hoping that adequate separation from all interfering peaks can be achieved for all analytes for all matrices in a reasonable time frame.

or

2 - Running standard addition on every sample

Am I viewing this correctly? Are there other options beyond the two listed above? Or is there a valid argument to continue to run external standards in water for this multiple matrix situation as I don't see a scenario where that could be called a good practice going forward.

If I'm reading your situation correctly, I don't think that standard addition will help. Assuming "coeluting" interferences (i.e., close enough that they can't be teased out from the analyte), unless you have a matrix blank standard addition will still give you a value that represents analyte + interference.

In fact you may have to go to a different column (Aminex or Rezex...) which means a different mobile phase, and a more sensitive detector like ELSD. Per ICH Q3 (pharmaceutical I know) you can tolerate a coeluting peak of 0.05% wt/wt of your target (API's Maximum Daily Dosage).

Thanks for the responses -

Tom - I was under the impression that standard addition was a way to account for matrix interferences including from co-eluters and interference from peaks that might not have good resolution from the analyte ie less then 1.5. And now I know I was wrong. So thank you for pointing that out and informing me of my errant thinking on that.

While we are not sure that there are actual "buried" co-eluters I do have matrices where large shoulders and even smaller peaks that are not fully resolved from the analyte are present that are not present in the aqueous standards.

Going to a new detector is not a possibility. Not sure how to deal with the interferences caused by the various matrices now.

Sorry about the delay in responding (I was hoping someone else would chime in with a solution)

The "best" (albeit not necessarily the most practical) solution is to improve the resolution between your analyte(s) and the matrix peaks. One move in this direction would be a more efficient column (higher plate number). That would generally come under the heading of "adjustment" rather than "modification" and so would not require revalidation.

The most practical solution might be to simply ignore the presence of those underlying peaks. If you're not pushing down to super low levels and/or if you don't need super high precision, it may not matter. As a variant on that, if you have at least *some* resolution between analyte and matrix, you might be able to improve things by fine-tuning your integration settings to carve off shoulders or "tails".

HPLC Chemist's suggestions are both good (but painful in that they definitely require validation). Changing to a different type of column with a different mechanism (IMP / anion exchange / HILIC) may still let you separate all your sugars on one run. Assuming you aren't pushing the limit of detection with RI, ELSD probably wouldn't do much unless your sugars are substantially less volatile than the junk.

Sorry to not be helpful, but sometimes there just are no easy answers.

If HILIC with an RID doesn't work you may have to go back to 'the bad old days' when the only option for carbohydrate analysis was derivatization (OPA or TMS?) and GC.

Hi Gizmo,

The standard addition couldn't work in such a case since you have no idea of what you have under your peak and how much of this "thing" is hidden below the peak of interest in each sample... adding a reference will not solve the unknow below the peak.

Tom is right: if you're not looking to small amounts, is it manadatory to go deeper on that matter ?

If you want to change your method, there are plenty of options today : HILIC, carbohydrate dedicated columns, amino-based columns (take care about water stability of the column).

One addition about the detection (!!!Bias alert : I am involved in ELSD manufacturing!!!), changing detection will not solve your issue, however evaporative detectors are far more sensitive than RID, in addition you can consider gradient of elution This is really true with carbohydrates that are all non volatile and make no difficulty for those detectors. So if you're looking for more information from your sample, that could be also an option. Check the evaporative detectors manufacturers websites, there are plenty of applications that could be a good starting point for you.

GC is, from my point of view the less desirable option, changing equipment and adding a derivatization step (yield ? pre-treatment ?) is a bit outdated considering the options you still have in HPLC.

I have some experience on carbohydrates (only with ELSD), if I can help...

Thanks again for all of the replies. Actually this problem is not quite as severe as I first thought. Our recent software upgrade allowed us to do overlays of the chromatograms and we did not realize that the option we were under modified the scaling to show the peak heights as equal.

This caused some confusion because the areas were vastly different and the only thing that made sense was the lower portion of the peak on the smaller sample seemed to be cut out by the integration and the effect appeared very exaggerated in the overlay display. And so after nearly being driven mad by this we figured out what was going on and found out how to do a proper overlay showing the real sizes of the peaks from each run relative to each other.

And the effect is not nearly as significant as it seemed under the distorted overlay so we will be proceeding as we have been prior to this incident while being more mindful of how the integration goes for each sample and particularly so when dealing with a new matrix.

For the record, for the future, another way to look at sugars is capillary electrophoresis. Beckman Coulter used to do a kit for the purpose, which derivatised sugars with a fluorescent phenolic thing bristling in sulfonyl groups to give it a reliable uniform charge. It's completely inappropriate for this thread because it means new methods on new instrumentation, but it's worth a mention because it was extremely sensitive and had very high resolution. I met it very briefly some years ago before the laser used in the laser-induced fluorescence detector failed, and the instrument moved to another lab. Its resolution made GC look clumsy and crude. There were all sorts of things in our samples of which we'd been blissfully unaware until CE arrived.
Unfortunately there is a very big barrier to chromatographers in using CE because the hardware is so different - it's a foreign world - but it is an excellent technique.