Automated optimization of chromatographic separations

The development of a chromatographic procedure for an unknown sample (mixture) requires the selection of a variety of experimental conditions (type and composition of the mobile phase, characteristics of the column and the stationary phase, temperature, flow-rate, pressure, type of gradient, etc.). This problem was traditionally solved in an empirical way and with the aid of the vast literature on similar situations already dealt with. The last few years have seen attempts at the rationalization and automation of the optimization of chromatographic processes which have resulted In Interesting systematic approaches of great use. The monographs by Berridge [56], devoted to HPLC, and that by Shoenmakers [57], which deals with both HPLC and GC, represent the most systematic and complete compilations in this field at present. 

As a rule, chromatographic optimizations are based on the trial-and-error approach, relying on experimentation, and the basic relationships of chromatographic theory. In HPLC [58-62] is used the well-known expression for the resolution 

The procedure followed in GC is similar, but the column temperature plays a role similar to that of the mobile phase composition in HPLC [63], 

The optimization of a chromatographic process usually calls for a large number of experiments which generate a vast amount of data. If it is performed manually (l.e. with the sole intervention of the operator), the procedure is slow and subject to major errors of interpretation, which may shift the search for the optimal conditions in wrong directions and call for an unnecessarily large number of experiments. 

The use of computers for optimization purposes has brought about considerable advances as they allow  

---