Limit of detection (revision 5)

The limit of detection is the lowest concentration of an analyte that can be reliably distinguished from background, though not necessarily measured with acceptable precision. The limit of quantitation, at which a number can be reported, is higher — conventionally by a factor of about three.^[1]

The distinction is practical rather than pedantic. Between the two limits an analyte is detectable but not measurable, and a determination reported in that region carries an uncertainty comparable with the value itself.^[1]

For peptide impurity work the relevant question is usually not what the limit is in absolute terms but whether it is below the reporting threshold. An impurity present at 0.05% is irrelevant if the method cannot see below 0.5%, and a certificate reporting "no impurities detected" is a statement about the method as much as about the material.^[2]

Three approaches are recognised. The signal-to-noise approach compares the analyte signal with baseline noise and takes a ratio of about 3:1 for detection and 10:1 for quantitation; it is simple and is the usual choice for chromatographic methods. The standard-deviation-of-the-blank approach uses replicate blank measurements. The calibration-curve approach uses the residual standard deviation of the regression and its slope.^[1]

The three do not necessarily agree, and a reported limit should state which was used. Signal-to-noise figures in particular depend on how noise was measured — over what region of the baseline, and after what smoothing — and are not comparable between laboratories without that detail.

The limit is a property of the method as applied, not of the technique. The same instrument with a longer injection, a more concentrated preparation or a more sensitive detector setting has a different limit.^[3]