Standard addition is a general approach in many techniques to make the analysis of complex samples

more robust, and also more accurate.

Typically, a measurement needs to be calibrated in some standard solutions. Then the unknown samples
are measured in the same way. Using the calibration the concentration(s) of interest can be determined.

This simple approach is less satisfactory when complex samples are analyzed such as body fluids. This is
because usually simple background solutions are used for calibration such as saline or PBS, spiked with
known amounts of the substances to be measured in the actual samples.

This does not work well for complex specimens because is impossible to prepare standard solutions
where the background solution is the same or similar to the background sample composition of complex
fluids. Therefore the calibrations will be less accurate since they are done in very difference solution
milieu than the real samples.

This problem is overcome to a large extent by using standard additions. Here a small volume of high
concentration of the substance in a simple buffer (like saline or PBS) is added several times to the actual
sample. This way very little dilution of the specimen occurs and thus its original background composition
is nearly preserved.

The question is how can the unknown concentration be determined from data measured after each
standard addition. The equations below will make this clear, using the example of spectrophotometry.

V volume

c concentration of the substance to be measured (M = mol/L)

x unknown specimen (subscript)

st added standard solution (subscript)

Beers law: A=adc

Here A is measured, d is known with a given cuvette, and a is theoretically known but practically it
contains unknown interferences from the complex background solution and also, non-idealities of the
instrument. This is why we need calibration. In some cases a bias term, b, also needs to be added to
Beers law if it cannot be perfectly zeroed at c=0, but inclusion of this is straightforward when needed.
Below the product a d is substituted with k.

Unknown sample:

Ax = k cx

1st addition:

A1 = k [ (Vx cx + Vst cst ] / (Vx + Vst)

2nd addition:

A2 = k [ Vx cx + 2 Vst cst ] / (Vx + 2 Vst)

A single addition is enough to determine the two unknowns: k and cx. A further addition makes the set
of equations obviously contradicting (for any real measurement error) but statistical methods (least
squares or regression analysis) can be used to determine the most likely solution which will be better
than what you get from only one addition, since it compensates also for some of the measurement
errors.

Further additions further improve the quality of the results. It is important, however, to keep Vx >> Vst
and cx<< cst.