MULTIVARIATE QUALITY ASSESSMENT OF AGED RP-HPLC COLUMNS

The inner structure of RP-HPLC columns can change ('age') during routi ne use. The columns may be monitored to check the magnitude and type o f the changes. One approach is to investigate, besides the chromatogra ms, the retention behaviour of a few selected solutes at different mob ile phases (markers) during routine use. A distinction can be made bet ween a general shift in retention and changes in selectivity. The chan ges in retention of selected solutes on aged columns in comparison wit h fresh columns are collected in one multivariate statistic, Hotelling 's T-2-statistic, instead of separate univariate statistics. The T-2-v alues are denoted on a control chart to visualize the changes in reten tion and to compare them with a control limit. Changes of T-2-values a bove this limit cannot be considered negligible. An alternative approa ch is to select underlying variables or principal components of a larg er set of solutes. These underlying variables can be treated just like the selected original variables. Examples are given of control charts for artificially aged columns. The results corroborate a global categ orization made of the artificially aged columns based on the retention times of a large number of solutes at different mobile phases. (C) 19 96 by John Wiley & Sons, Ltd.