Optimization of coal tar gas chromatography conditions using probabilistic-deterministic design of experiment

The development of physicochemical methods for the analysis of objects of complex composition requires the application of methods of mathematical experiment design. This article investigates the possibility of using probabilistic-deterministic design of experiment (PDDoE) for obtaining a mathematical model of the chromatographic separation process of coal tar hydrogenation products on an Agilent 7890A device with an Rxi-5ms column. It is shown that the relationship between the column heating rate and the carrier gas pressure with the values averaged for the entire chromatogram can be established with a high accuracy. It is noted that the accuracy of modeling the individual characteristics of the mixture components’ peaks is less, but remains sufficient for many practical needs. Nonlinear multiple correlation coefficients (NMC) for the dependence of the average retention time and average intensity on the considered factors are more than 0.99; they are more than 0.98 for the average peak width. NMC for the dependence of the resolution with the relation to the peaks of naphthalene and 2-ethylphenol is more than 0.8 at a significant level that sufficient for practice. The quality of the mathematical model was checked by triple registration of the chromatogram at the values of the column heating rate and carrier gas pressure that were not used in the training experiment. The measurement results are excellent squared with those calculated using the obtained generalized equations. The PDDoE method can be recommended as a method for mathematical design of an optimization experiment in gas chromatography.