Defining an SEC calibration

Most commercially available data handling systems which have SEC software written for them assume that the calibration will be carried out using a series of narrow MMD calibrants. The molecular masses are always used in the logarithmic form, and the calibration is usually established either as a linear relationship between the log molecular mass and the elution time, or by using a polynomial relationship. Different systems permit the use of various orders of polynomial, but in practice a linear fit or a third-order polynomial are the most commonly encountered. 

If a linear plot of elution time against the log molecular mass is a good description of the elution behaviour of the calibrants, then this is a very satisfactory manner of defining the calibration. The manufacturers of SEC columns now put a significant emphasis on producing columns, or combinations of columns, which can effectively be defined by such a linear plot. However, if the linear definition does not accurately reflect the observed elution behaviour of the calibrants, serious errors in the calculation can occur. 

A third-order polynomial, of the log molecular mass against the elution time, frequently appears to be an accurate definition of the calibration. Unfortunately, a third-order polynomial is not ideal for defining the calibration in practice it is very difficult to reproduce such a calibration, with quite small differences at the low-molecular-mass end causing large changes at the high-molecular mass end of the calibration. When these discrepancies are observed as the actual numbers, rather than their log value, the differences can appear to be very large indeed. 

Alternatively, the calibration can be described by a multi-straight-line plot, a series of linear calibrations between each point. This manner of describing the calibration can be quite accurate and reproducible providing that a large number of points are used. Unfortunately, if the calculation is carried out 

---