coeluting or comigrating impurities

[ntruong]

Hi All,
I have been trying to develop an HPLC method for a weak organic base compound. I have no trouble spotting the peak of interest using either Gradient or isocratic (mobile phases: 10mM AMA pH 3.0 and ACN) with columns 1) C18 Supelcosil LC-18-DB; 2) C18 YMC J'sphere and 3) Agilent Zorbax XDB-C8. However, the peak purity for 1, 2 and 3 passed at 0.25mg/ml concentration but fail for higher concentration 0.4mg/ml +(I have also tried to maintain the concentration below spectra absorbance of 1000mAU. I then switched to Phenomenex Gemini C18 (mobile phases: 10mM TEA, pH 11.2 and ACN). Same result. The UV spectra profile of the peak of interest are the same, the peak shape is symmetrical and tailing is about 1.1 - 1.3. Could there be coeluting peaks? How can I separate them, if present? Is there other mobile phase system I can use to separate them?
Thanks for your help,

[tom jupille]

I have two words of advice: "orthogonal separation". All of your C18 columns seem to provide similar selectivity. You basically need to find a system with different selectivity. The variables you have to play with are:

gradient steepness
temperature
organic solvent type
pH
additive concentration
column chemistry

It's your choice as to which variable (or combination) you want to play with. If it were my problem, I'd start with your favorite of the C18 columns and run a couple of gradients with different times (maybe double and half what you used before). See if you can pull any additional peaks out. Next on my list would be temperature (35 and 50 degrees ?), then organic solvent type (switch from ACN to MeOH). You've already looked at pH and additives and saw no change. If you want to exploit column chemistry, you want a packing that's as different from C18 as you can get. Generally, this means one of the "embedded polar group" columns.

Good luck!

[ntruong]

Hi Tom,
For the past couple of days, I tried to fiddle with the variables as you suggested (gradient, temperature, solvent) and I still couldn't push the invisible peaks that buried underneath out. It bothered me when I tried to splice it using the peak spectrum tool and there is no indication of peak overlapped since the UV profile is the same. However, when I started using the peak purity tool (Agilent DAD), it failed. I could see peaks exceeded the threshold. By the way, what is the magic number of threshold? of noise deviation? Do you have any embedded polar type column in mind that I could use. I don't know what I have in stock, but I will try to find one. Is it phenyl?
Thanks,

[Rafael Chust]

Since you referred Phenomenex Gemini C18, you might try to use Phenomenex Synergi Max-RP or Fusion-RP, that although works quite like a normal C18 column, it will enhance the differential selectivity Tom referred.

Phenyl works quite well if your analytes have aromatic rings, since Phenyl enhances the π-π interactions.

[bkirschil]

Ntruong,

It looks like pH manipulation will not work since you tried acidic and alkaline mobile phases with no apparent different. I would suggest changing the column chemistry based on the functional group analysis of the analyte. If by PDA you can see the UV spectra of one or more impurities, try to determine approx. similarity in structure and exploit differences. I would now try columns with smaller particle size and perhaps a column that can handle both polar and nonpolar species (there are several). You may also want to switch to a smaller column with smaller ID and particle size under 2-um (e.g., Waters Hybrid). AS I said analyze the analyte chemistry and functional groups well before selecting a more appropriate bonded phase.

[Kostas Petritis]

That is where mass spectrometry could be very handy...

But back on the problem, I would try to introduce some ion-exchange capabilities in the column either through ion-pairing chromatography, or by using some mixed chemistry columns or Hypercarb in high pH or HILIC...

[HW Mueller]

I just wonder whether the spectrum could change at such high concentrations? I don´t remember, are nonideal (deviations from Beer-Lambert) absorbances the same at all wavelength?

[Bryan Evans]

Our 3 micron Unison columns provide a different selectivity
from regular C18 columns. In addition, they get
an incredible 200,000 plates/m.

http://www.silvertonesciences.com/index ... var=LINK_1

[Rob Burgess]

I'm guessing your peaks are probably pure but its the manipulation of the software peak purity parameters that is making your peaks appear "im-pure". What peak purity software are you using?
We recently had a situation very similar to your own... a low level column loading appeared pure, whereas increased column loading appeared not to be. Tracking through how to use the software from a very useful internal SOP enabaled us to "fudge / tweak" (take your pick of words here) the software to read the analytes as "pure".
Nevertheless Tom did make a good point of using an orthogonal separation technique (or MS peak purity) as a double / double check on peak purity!

[Scottie]

The previous post is probably correct. I see you are you using the Agilent ChemStation software, so the 2 parameters to check first are the wavelength range you are using and the threshold value.

You should check the wavelength range you are measuring the peak purity first. You may find that although the wavelength of interest is below 1 Au, other wavelenghts are above 1 Au. Adjust the wavelength range options in spectra options to only include wavelengths below 1Au.

Next is the peak threshold. This is harder to set and will vary from compound to compound or even from instrument to instrument. It is a comparison of the spectra at time x across the peak vs. the mean spectra of the peak. Do not assume that this is comparable to the % purity of the peak or compound.

ChemStation calculates what it thinks the the threshold should be based on the baseline noise at the begining of the chromatogram, but it can be manually altered.

Where possible I set this using an injection of a fully characterised (reference) material. This I assume to be pure, and use the purity value of the reference peak to set a baseline value. Be careful not to set this limit too tight though. For example if the peak purity of the reference is 995, I would set 990 as my threshold.

I hope this helps

Mike

[ntruong]

Hi Scott,
The peak of interest is maxed at 306nm. I set the range at 245nm to 500nm for peak purity determination. The threshold is at 1, reference at 990, noise deviation at 0.1. Do they look reasonable to you?
thanks,

[Peter Apps]

"The UV spectra profile of the peak of interest are the same, the peak shape is symmetrical and tailing is about 1.1 - 1.3. Could there be coeluting peaks?"

A uniform UV spectrum and a symmetrical peak does not rule out the possibility of co-elutions, but the "fail" on the peak purity must be spurious because it is departures from spectrum uniformity across the peak that the peak purity function is looking for.

It is a while since I used HPLC but I recall that with the purity criteria set too rigorously any peak would fail.

If the peak remains UV uniform with the changes that you have made it is very likely to be a single compound.

Regards Peter

[JM]

Increase the noise deviation till you get the peak "pure".

it is as simple as that

JM

[ntruong]

Hi JM,
I did as you suggested and it passed. However, is there a limit to that? It is kind of defeating the purpose when you fiddle to the point the peak has to be spectrally pure. If possible, can you or any fellow chromatographer provide me a set of limit of peak purity settings that is widely used by pharmaceutical industry? By the way, do we care about peak purity at all? Is it a must for LC method development? What happen if you work with the GC method? How can you tell if there are no features available for you to play with? How can you tell the peak under GC conditions is pure?
Thanks in advance,

[HW Mueller]

This has been elucidated many times: To be sure about chromatographic results you need to perform at least one other type of analysis, a widely differing chromatographic method might do also. UV doesn´t help much (too little uniqueness) high resolution MS is quite good but not perfect. That´s why organic chemists use even elemental analysis quite a lot yet, besides chrom., nmr, IR, MS, Raman, UV/VIS, fluorescence....