Saturation of Absorbance Detectors? How much is too much?

[dkreller]

I am doing a weird variation of HPLC right now in which I am injecting quite large amounts of sample onto a column. I am actually studying adsorption; I pack my own columns with adsorbents that bind components of the sample (mixture). Because I am collecting fractions of column effluent I inject large amounts of sample - to get a lot out to facilitate my other analyses. However, I am caught between a rock and a hard place. As you'll see in the image, because I am injecting large amounts and want to saturate the adsorbents in the column in a reasonable number of injections, I tend to saturate the absorbance detection channels.

My question is this: At what value of absorbance do you figure the detector is saturated? I know that when you use Beer's law in any other application you say keep the absorbance below 1. Should the same rule of thumb be applied to absorbance detectors? Or, do they have some special ability to measure higher absorbances? I feel the the data with absorbance higher than about 1 shouldn't be entirely trusted. (I use peak areas to calculate amounts adsorbed). However, I note that even the high absorbance later injections still keep the expected peak shape, and don't really show distortion or clipping. I am interested in hearing the thoughts of a few good folks on this forum.

Thanks,

Dave .

[aldehyde]

What kind of detector are you using? You may have to study linearity on your own, but for common instruments it should be published.

http://www.chem.agilent.com/Library/tec ... 1890EN.pdf

[dkreller]

Sorry that I was completely negligent; I didn't say what kind of detector. I have a system built entirely of Waters components. The detector is question is a 2487 dual-wavelength detector.

Dave

[lmh]

Waters should know the real answer, but most modern instruments can do better than 1; If you're working with absorbances between 1 and 2, you're working between 10% and 1% transmission, which doesn't seem an impossible demand on a modern instrument.

Of course there's nothing to stop you picking a wavelength at which your analyte only absorbs slightly, and using this for the more abundant components.

[JGK]

Being an old traditionalist I generally work at under 1Au full scale.

Although the waters 2487 is capable of working between 0 and 4 AU, Linearity and reproducibility will be a challenge.

[HPLCaddict]

Mr. Troubleshooting also had something to say regarding this topic:

http://www.sepscience.com/Techniques/LC ... Conditions

[lmh]

Normally I'd take anything John Dolan writes as 100% truth to be followed in all circumstances, but I'm a bit reticent about this particular article: it needs to be treated with caution.

He starts with the assumption that detectors are not linear beyond an OD of 1 unit, and on that basis makes up some data with an arbitrary non-linearity of 5% at 1.5 and 10% at 2.0; he then forces a straight line through the result and demonstrates that it gives a bad estimate of calibrated amount.

The article tests the consequences of slight deviation from linearity (which is a good thing to do). It doesn't, however, address whether the deviation from linearity exists. That's down to experimental evidence; our decision about what OD we can work at should be taken based on the genuine performance of our detector.

Incidentally, I'm also dubious about forcing straight lines through data that are known not to lie on a straight line.

[dkreller]

Thanks for your input folks. I can imagine that potential non-linearity in detector data and the concomitant effect on results of quantitative analysis is very serious business for folks doing quality control work and other analyses. This work of mine is only semi-quantitative. Working at these high values of absorbance I have still been able to extract useful trends from the data. However, if I try to publish this work, I will definitely address the potential effects of non-linearity in detector response. I plan to do a set of test injections in which I systematically increase the amount of organic matter injected up to the largest injection I can make, and I will test the areas under peaks for linearity/non-linearity. I will follow up by posting the results.

Thanks,

Dave

[tom jupille]

The reason that a lot people (including me) are uncomfortable with absorbances greater than 1 is that UV detectors actually measure transmittance; the absorbance is a calculate value (the negative log of the transmittance). It's impossible to make a perfectly light-tight detector so the you can reach will ultimately be limited by stray light. If you have 1% stray light (i.e., the stray light is 1% of the intensity of the light from your source), then you will *never* see an absorbance > 2 and will, of course, see non-linearity well below that.

That said, we can speculate all we want but I think we all agree that *you* have to determine the linear range on *your* detector.