Effect of Wavelength on Peak Width?

[adam]

I have been doing chromatography for many years but I recently observed something that I cannot make any sense out of.

We are running a method at 2 wavelengths: 310 and 220. The API gives a peak at 310 and the antioxidant gives a peak at 220. The strange thing is that the API also shows a peak at 220 but it is wider than the peak that we see at 310. Does this make any sense?

One point to mention: at 310 the API peak is at a maxima in the UV spectra while at 220 the API has a steep slope in the UV spectra. Would using a wavelength that is on a steep slope somehow lead to a peak that appears wider?

I would like to better understand what is going on and appreciate any suggestions.

[reyoungs]

Adam,

Peak width is a function of the the absorbance sensitivity of the chromophore. (Molar extinction coefficient or absorbance value in Beer's Law) You don't mention reltive intensityof the absorbance at each wavelength. I don't htink it is the slope but rather the absorbance value itself.

Run a spectrum of the analyte on a static UV or PDA. My guess is that your compound is very much more sensitive at the end absorption region (220) than it is at 310 nm and you are measuring on a rapidly rising upslope. As a result, your sensitivity or QL will change proportionately and you will see presence of trace amounts absorbed on the column at 220 which are not detected at 310. We forget that the absorbance term in Beer's Law is unique to each wavelength.

Most compounds have very strong absorbance at low wavelengths. You trade off sensitivity for selectivity when you select a higher wavelength. If the extinction coefficients are signficantly different, then the area under the curve will increase proportionately and the width probably will also, depending on column loading and particle size.

I think once you run a spectrum, you will find that this behavior correlates with absorbance and you are simply seeing much more sensitively at the 220 nm wavelength.

You may also consider the bandpass of the detector as well. Most older instruments had wide bandpasses, meaning that even if you set the wavelength at 220, the sample would see everything from 215 - 225. This makes the above sensitivity discussion even worse when you are including wavelengths from very sensititive regions and comparing them to something that is essentially a plateau at the maximum wavelength.

Let me know if this is the case.

Regards,

[mohan_2008]

Reyoungs has given an excellent explanation.

On a steep slope (220 nm), the extinction coefficient changes steeply, and this effect is amplified if your set bandwidth is larger. This would give rise to a hump/broad peak even though it is a very sensitive wavelength.

On a plateau (310 nm), the extinction coefficient changes very gradually - hence will not be amplifed as much, using a broader band pass setting.

Hope it helps.

[tom jupille]

I think the preceding posts focused on spectral peak width.

The width of a chromatographic peak should be independent of the wavelength. That said, it is difficult to accurately measure the width of a very large (overloaded) peak. If you can keep everything on-scale, the width at half-height it a more reliable measure than the baseline width in this regard.

If the width at half height is indeed greater at 220nm, this suggests that you have something else coeluting with the API. Almost any compound with a double bond will show at least some absorbance down in that range.

[HW Mueller]

In other words, if you can overlap the peaks such that they have the same hight , they better overlap exactly, everywhere. The reason for any discrepancies is given by Tom, I can´t think of any other ones.

[JayVee]

There is also the possibility your API is coeluting with another compound which absorbs at 220nm, but not 310 (like your antioxidant, if I understood your post well).
Is your API peak fronting/tailing? If coelution is the problem, you could be looking at a degradation of your API during the chromatography.

Do you have a DAD in your lab by any chance, or only dual wavelength detectors. A DAD should be able to help you in determining whether it is the case or not.

It don't think it is a change in your API with pH, because in that case you would get only one peak (with changing RT of your pH is poorly chosen), even if your API is present in two forms, and that peak should have the same width at all wavelengths.

[Uwe Neue]

If your API is completely out of the linear range of the detector at 220 nm, it is possible that it looks wider. Essentially the top of the peak gets a lower response than the bottom. You would see this from a claibration curve.

[adam]

I think Uwe got it.

I went back and looked at it, and I think the upper part of the peak is showing less absorption relative to the lower part. It has a somewhat atypical peak shape that is consistent with this. And, hence, the peak 'appears' fatter. The area count for the peak is around 33,000 and I think this is often where things start to go out of range.

I also like HW Mueller's statement that - absent any artifact like we've seen here - 'if the height of the peaks is normalized (for two different wavelengths) the two peaks should superimpose on one another.'

Much Thanks for all the feedback.

[mohan_2008]

Hi JayVee,

How can we use DAD as a tool in detecting "co-elution problems" and Impurity identification.

[JayVee]

There should be a function called "peak purity". You will need to look at your instrument manual for more details.

But a simple visual examination of the 3D chromatogram should give you enough information: if there is absorbtion at wavelengths where your API isn't meant to absorb at or near the retention time of your API, that means you have coelution of something else.

From memory, peak purity compares the ratio of the absorption at different wavelength for different timepoints for a certain duration, i.e., is the absorption spectrum the same with time, but simply varying in intensity (due to variation of the analyte concentration in the detector cell), or is it the superimposition of two (or more) absoption spectra (coelution) each varying in intensity.

[reyoungs]

Adam,

Uwe and I are discussing the same thing from different perspectives. If the peak absorbs intensely, then you will see overloading of the detector and possibly the column as well. Wider instrument bandpass makes the situation worse, but if you have overloaded your detector, then the peak will appear wider and have an irregular, non-normal top. The higher the absorbance value, the more you can see traces of your analyte, also making your peak wider and also increasing the possibility that you are saturating your detector.

Tom,

I don't see how the spectral bandwidth of the analyte impact this issue as much as the absorbance values observed at any given wavelength. Please refresh me on this.

[tom jupille]

I don't see how the spectral bandwidth of the analyte impact this issue as much as the absorbance values observed at any given wavelength.

Rick, I don't think it does (and I hope I didn't imply anything to the contrary!). In the absence of overload or a coeluting impurity, the peak width should be independent of the measurement technique (okay, I'm also ignoring extra-column volume here); i.e., what's coming out of the column is what's coming out of the column.

The consensus of the thread seems to be settling on overload as the problem. In that case, I can imagine that either increasing the bandwidth or shifting to a lower-absorbing wavelength might help by alleviating the spectrophotometric overload (assuming that we're centered on λmax, increasing the bandwidth would result in a larger percentage of the total light flux coming from lower-absorbing wavelengths). I agree that increasing the bandwidth would have less of an effect than shifting the wavelength, but it might also pose less of the kind of reproducibility problem you can get if you're working on a steep absorbance slope.

[danko]

I agree with Tom – completely.

Just to refresh some facts:
1. Peak width in chromatography is related to time and not to absorption coefficient.
2. The only overload that can cause peak width increase is mass overload and not detect - or/ion overload.

Best Regards

[Uwe Neue]

Any non-linear detector will give false plate counts. If the non-linearity is a downward curvature the tip of the peak will be suppressed. With other words, the base will appear to be wider than for a normal peak.

In the old times, you could see the nonlinearity of the detector from strange dips and and noise at the top of the peak. That triggered your brain to tell you that you overloaded the detector. With modern detectors, I have not seen such a thing.

In addition, the absorbance in the low UV is in many cases several fold higher than that at higher wavelength, which again would speak for my interpretation.

We will know for sure what the problem is if Adam finds the time to do a calibration curve at both wavelengths.