Indirect UV detection

<p><span>Has anyone had success using UPLC and indirect UV detection for analytes (< 250 Da) that do not have a UV chromophore?</span></p>


  • Dear Jim:

    What do you mean more specifically by indirect UV?

    As you know we have not created UPLC versions all of existing HPLC detectors. There was a request for RI on this site too!

    I will be interested to see what others contribute to this question.


    Edited by: Elizabeth Hodgdon on Dec 3, 2008

  • Greetings Jim,

    First off, I have never attempted a separation based on indirect UV detection. However, I do understand the principles behind the technique. For those peeking (small pun) in on this discussion who are unfamiliar with the term "indirect UV detection", it is a technique whereby the mobile phase, or eluting solvents, intentionally include a UV absorbing compound. This creates an elevated UV background that can be "displaced" by an eluting peak that has no UV absorbance. When this happens, a negative peak, rougly proportional to the concentration of the peak, is generated. At one time or another we have all experienced this phenomenon. It is the primary cause of the negative peaks that are often observed at the void volume in a chromatogram. Non-absorbing components in the sample diluent (water) displace the small amount of UV background in the mobile phase and a negative peak is generated. Back in the day, when recycling isocratic mobile phases was more widely practiced, it wasn't unusual to start generating an entirely negative chromatogram. As the injected compounds began to "enrich" the mobile phase, their concentration would, eventually, be greater than the concentration of the subsequent sample injections. The auto-zero function on the UV detector would mask and compensate for the, constantly, rising baseline. So, the peaks would, first, "shrink" and then "grow" into a "mirror image" chromatogram.

    Which brings me around to the couple of things to be mindful of if this technique is attempted. Most instrument manufacturers (Waters included) pretty much assume that the peaks of interest will be predominently "positive". Most detectors (RI excepted) and data systems make a small allowance for negative peaks. The response regions (the analog to digital converters) of the detectors and data systems are almost always biased towards the positive. In most of our detectors only 10% or 20% of the dynamic range is devoted to the negative. If a negative peak exceeds this margin it will, essentially, saturate the electronics and square off or "flat top" (in this case, "flat bottom"). This is exactly the same behavior as you would see if your peak had an absorbance of, say, 6 AU. It would skyrocket to the heavens, go completely flat (horizontal) and return to baseline sometime much later than you had hoped. This would suggest to the, would be, practitioner of "indirect UV detection" that turning off the "auto zero on injection" function of the UV detector would be a good idea. If this function is left "ON" (usually the default), the UV absorbing mobile phase, sitting at an absorbance of something like 0.5 to 2.0 AU, will be, automatically, turned into zero. This would, once again, leave you with only the 20% dynamic range of the detector with which to work. Once you have sorted out the various dynamic range issues, the rest would be reasonably straight forward. Most detectors and/or data systems allow the user to "reverse polarity" in some fashion or another. However, a word of caution here. This reversal does not, usually, reverse the dynamic range bias. It would simply place the 20% in the "positive" direction rather than give the "negative" (now positive) the 80% of the dynamic range. For those a little more adventurous and not prone to vertigo, many data systems (Empower for one) are more than happy to integrate negative peaks which would render all of the polarity calisthenics unnecessary. You would simply have to get used to "up" being "down".

    Sorry for all the words. Hope this is useful.


  • Jim

    Hi Liz,
    Indirect UV is when you perform HPLC with UV detection and you are typically trying to detect analytes that do not have a UV chromophore. The separation
    mechanism can be reversed-phase with an additive (e.g. KHP) that gives the mobile a moderate UV response. When the analytes elute the UV background
    decreases and a negative peak is generated. The detector output is reversed so that the negative peaks turn into positive peaks. An example of this is to
    detect common anions (e.g. Cl, SO4, Br, etc.) using an ion-pair mechanism with TBAOH. You can perform an anion analysis on your HPLC/UV system. We did this
    years ago. We now are interested in looking at some small cationic analytes using the same concept. I wonder if it is easy to reverse the out put of the
    PDA? Do you know this?

    I am asking this question to see if others have come up with new additives for indirect UV.


  • Hello Jim,

    PCB has made some good points regarding the expectation on the part of UV detectors for the peaks to correlate positive signal with increased absorbance and impose some restrictions as a result. Unfortunately, the dynamic range of the PDA is probably even less than the 20% or so that PCB hoped for. The PDA has no other choice than to "autozero" at the start of a run by virtue of the fact that it captures an initial light spectrum as a reference, and then computes absorbance relative to it thereafter for the duration of the injection. Prior to that reference spectrum capture, a lamp optimization takes place, which is really an optimization of the electronics and software that detect and quantify the light level from the lamp. This essentially causes the light detection facilities to be "maxed out" with only a little overhead for some drift because, of course, the expectation is for a series of positive absorbance peaks (i.e. less light than the baseline).

    I would expect success with the PDA if the mobile phase was adjusted through gradient programming to be "clear" (non-absorbing) at the very beginning and very end of the chromatogram, which would cause it to be clear between the injections, and then stepped up to the higher absorbance level by introducing a higher ratio of the absorbing mobile phase well before the peaks of interest. The step gradient could take place as early as time = 0.00 because the reference spectrum is captured just prior to time = 0.00. After the last peak, clear mobile phase conditions could be restored. So, the negative peaks would be dropping down from an elevated plateau during the injection, but as PCB noted, Empower should have no problem integrating them.


    Brian E.

  • Greetings Brian,

    I was hoping you would offer some insight on this question. I wasn't exactly sure how much of the dynamic range was given over to negative responses. I suspected that 20% was probably generous. You are correct, I was thinking more of the TUV when I wrote my, original, comments. As you say, the PDA has little choice but to re-zero at the begining of each injection. I also wanted to add that it appears neither of the ACQUITY UV detectors offer a reverse polarity option on the Ethernet signal. There is still the ability to reverse polarity (switch wires) off the analog outputs and bring them into an analog to digital collection and conversion device such as the SAT/IN on Empower.



  • Jim

    Brian and Peyton,

    Thank you for your comments. I have had success! I have both PDA and TUV detectors and have been using the TUV. I have been using the analog out signal since I can reverse the polarity. This signal is sent to my analog input on my ZMD which is controlled through MassLynx. In the future can we have access to controlling the auto zero function and polarity on the TUV through the non-analog signal (i.e. direct signal to MassLynx)?