Non-linearity of calibration in the determination of anions by ion-chromatography with suppressed conductivity detection

Traces of simple inorganic anions may be determined by chromatographic separation with an alkaline eluent and conductimetric detection, which involves “suppressing” the background conductivity of the eluent by neutralization to form a sparingly dissociated species. Over a wide range of determinand concentration, e.g., two decades, non-linearity of the calibration may become evident, leading to errors of up to 100% at lower concentrations if linearity is assumed (a linear fit of the data usually gives a correlation coefficient>0.99, which may lead to false confidence). The curvature arises from displacement of eluent ions by the determinand and the consequent re-equilibration of the conjugate acid in the suppressed eluate. Even if the distribution of determinand in the peak is ideally Gaussian, the observed conductivity peak may be distorted and calibration will then be non-linear. The best linearity is obtained with the most strongly basic eluent, but other characteristics must also be considered, e.g., run time, peak separation. With a carbonate eluent, the curvature is demonstrated empirically for chloride, nitrate and sulphate calibrations. A second-order fit gives errors of < 10%. With a more strongly basic borate eluent, the deviation from linearity is negligible, but elution times are longer and may be inconvenient in some circumstances.