Laboratorio de Separaciones Analíticas, División Química Analítica, Facultad de Ciencias Exactas (UNLP), La Plata, Argentina.
J Sep Sci. 2012 Oct;35(20):2699-709. doi: 10.1002/jssc.201200197. Epub 2012 Sep 20.
The prediction capability of the solvation parameter model in reverse-phase liquid chromatography at different methanol-water mobile phase compositions and temperatures was investigated. By using a carefully selected set of solutes, the training set, linear relationships were established through regression equations between the logarithm of the solute retention factor, logk, and different solute parameters. The coefficients obtained in the regressions were used to create a general retention model able to predict retention in an octadecylsilica stationary phase at any temperature and methanol-water composition. The validity of the model was evaluated by using a different set (the test set) of 30 solutes of very diverse chemical nature. Predictions of logk values were obtained at two different combinations of temperature and mobile phase composition by using two different procedures: (i) by calculating the coefficients through a mathematical linear relationship in which the mobile phase composition and temperature are involved; (ii) by using a general equation, obtained by considering the previous results, in which only the experimental values of temperature and mobile phase composition are required. Predicted logk values were critically compared with the experimental values. Excellent results were obtained considering the diversity of the test set.
研究了在不同甲醇-水流动相组成和温度下反相液相色谱中溶剂参数模型的预测能力。通过使用精心选择的一组溶质作为训练集，通过回归方程建立了溶质保留因子对数 logk 与不同溶质参数之间的线性关系。回归中得到的系数用于创建一个通用的保留模型，能够预测在任何温度和甲醇-水组成下的十八烷基硅烷固定相的保留。通过使用具有非常不同化学性质的 30 种不同溶质的另一组（测试集）来评估模型的有效性。通过使用两种不同的程序在两种不同的温度和流动相组成组合下获得了 logk 值的预测：（i）通过计算涉及流动相组成和温度的数学线性关系中的系数；（ii）通过考虑先前结果获得的通用方程，其中仅需要实验温度和流动相组成的值。预测的 logk 值与实验值进行了严格比较。考虑到测试集的多样性，得到了非常好的结果。
