Conceptual: Response factors

<p>I'm relatively new to HPLC and response factors are giving me some trouble.  My supervisor believes that a response factor can be created between any two compounds that absorb at the same wavelength and do not interefere with eachother.  In practice, this does seem to be the case.  However, if any two compounds happen to have different slopes (Area vs Concentration) then a response factor can only be created at a specific concentration.  A universal response factor cannot be created that works at all concentrations, even if these concentrations are all in the linear portion of the curve.</p><p></p><p>In an attempt to prove this, I ran 0.01 mg/mL and 0.2 Vitamin A acetate in duplicate and ran 0.01 mg/mL and 0.2 Vitamin A palmitate in duplicate.  I received different response factors when comparing the two concentrations.  I have not tested the upper limit of Acetate in terms of linearity, so there does exist the possibility that this response factor is different for that reason.  However, at least in theory- if two compounds have different slopses, one cannot create a reponse factor between them, correct?</p><p></p><p>Thank you for your patience in a possibly "newb" question.</p><p>Wes</p>


  • RD

    There are detector's that come close to providing a universal reponse, Evaporative Light Scattering & Charged Aerosol Detectors, a UV absorbance detector is not one of of them.

    UV absorbance is governed by Beer's Law (or the Beer-Lambert Law) whose equation is:

    A = ebc.

    Where A = absorbance, e = molar absorbtivity of the sample, b is the path length of the sample (flow cell path length for an HPLC/UPLC UV Detector), and c is the sample concentration.  e is nothing more than the absorbance exhibited by a molar concentration of a substance at a defined wavelegth and defined sample path length.

    A quick internet search will provide you with more information about e than you could possibly want.

    Suffice it to say that different compounds have different e values.  Because of this, the same mass concentrations of 2 different substances will produce different absorbance responses (i.e., different response factors).

    However,  one would expect the response factors calculated at different concentrations of the same compound to be the same (i.e., define a linear relationship).

    Beer's Law is linear for a given wavelength of light.  What compromises linearity in practice is the effect of "stray light".  Stray light is light of unintended wavelengths which reaches the sample.  It is a normal consequence of imperfect optics.

    I would expect an ACQUITY TUV to be linear at absorbance values >=  2 AU.

    How different are the response factors you are observing ?

    What is the absorbance value you observe at the high concentration ?



  • Thank you very much for replying.  I was getting an Rf for Niacin vs niacinamide and at concentration 0.05 mg/mL the Rf was 1.67.  At 0.01 mg/mL the Rf was 1.09.  The AU range was 0.8 to 0.2.  So I don't think theres any problem there.  Since my post, I've experimented and read up some.  Beer's law never fits my calibration curves perfectly and thus I get something more like A=ebc+y.  This y-intercept component to the calibration curve is typically 1000 to 5000; I think this denotes a small amount of interference of some sort which you described as stray light.  The y-intercept causes the Rf problem at different concentrations.  What I've done is changed the processing method from "Linear" to "Linear thru zero" to force my calibration to remove the intercept.  I'm a little skeptical as to if forcing a 0,0 y-intercept is a ligitimate practice.  Doing so has kept a good R^2 (>0.999 so far) using 4 data points (including 0,0) however my lower concentration has a deviation increase of about 15% (in this case from 4% to 19%).  I have also since tested the resulting Rf and it so far has given sensible results (versus vendor CofAs).

    On a side note you mentioned linearity is good at >= 2 AU, did you mean <=2AU?  I've been told that anything above 1.0-1.5 AU is getting sketchy.

    Thanks again for the help!


  • RD

    Positive y-intercepts are not uncommon.  In your case, I doubt that this is an effect of stray light but perhaps something more mundane like carryover.  As long as the curve is linear, I wouldn't worry about the individual response factors.

    I personally never force calibration curves through zero but prefer that they accurately reflect the observed data.  Forcing through zero will obviously change the slope of your curve (unless your data system can subtract the y-intercept from each calibration point) and is a risky proposition for quantitative analysis.  If however you can demonstrate the validity of your approach, I would not insist upon it.

    The limit of linearity for HPLC UV detectors is typically considered ~ 1 AU.  However, this may be more a matter of history and tradition than an accurate reflection of the capabilities of state of the art equipment.  The published linearity spec of the ACQUITY TUV is < 5% deviation @ 2.5 AU for propyl paraben.  Of course, use your data as a guide.  If a peak reaches its apex as a "point" and is not rounded off or flat topped and the calibration curve does not flatten out at the highest concentration levels, you should feel confident with the data.