The calibration sampling paradox of process analytical chemistry

7 The calibration sampling paradox of process analytical chemistry 

In any chemometrics application, it is critical to obtain calibration samples that are both sufficiently relevant and sufficiently representative of the samples that the model will be applied to. Furthermore, for quantitative applications, the reference concentrations of the calibration samples must also be accurate and precise enough to result in a model that performs within desired specifications. In PAC, it is often difficult to obtain calibration samples that are both highly relevant and very accurate. For extracted process samples, 

Real process standards, but reference concentrations obtained from process sampling/lab analysis or temporary on-line method 

Real process material used for calibration standards, reference concentrations obtained from a process model 

There are many different ways in which the calibration strategy can be moved toward the favorable upper right comer of the relevance vs. accuracy plot. In most cases, these rely heavily on prior knowledge of the specific process and measurement system involved in the problem. Such prior knowledge can involve the process dynamics, process chemistry, process physics, analytical objectives, analyzer hardware, analyzer sampling system, and other factors. 

---