|| Checking for direct PDF access through Ovid
pH and the binary organic composition compatibility was evaluated in broad range.A grid form model predicted the retention behavior of the ionizable analyte.The retention prediction of the grid model was highly accurate.13 acidic and 16 basic analytes were separated at the optimized conditions.The visualized summary of separation was useful for robustness estimation.A novel computer-assisted methodology for the simultaneous optimization of aqueous pH and binary organic eluent composition through a broad range of analytical conditions of reversed-phase ultra high-performance liquid chromatography is proposed. Two of nonlinear prediction models were employed to fit into the retention time (tR) on a linear gradient elution with a predefined slope. One model was derived from Bernoulli-type probability distribution to predict the value of tR against the pH value of the aqueous eluent. This sigmoid-shaped model was successfully fitted for tR value shift in the presence of three levels of organic eluent compositions (volumetric mixing of acetonitrile/methanol ratios 1:0, 1:1, and 0:1). The resultant pH versus tR value models were subsequently combined into grid form by quadratic multiple regression models based on the solubility parameter theory and their binary organic composition axes. The predicted tR values afforded from grid models were highly accurate for 13 different acidic non-steroidal anti-inflammatory drugs [root mean square error (RMSE) ≤0.030] and 16 basic histamine H1-receptor blockers (RMSE ≤0.067) in a pH ranging from 2.5 to 9.0 and an acetonitrile/methanol volumetric mixing ratio ranging from 1:0 to 0:1. Each compatibility score was defined as the indicator of the peak separation. Scores were calculated for all combinations of aqueous pH values and binary organic compositions via the predicted tR values. A colored map generated from the calculated scores was greatly effective in determining optimal combinations of both mobile phase conditions. By employing this predictive data, all analytes in both acidic and basic sample mixtures were finally separated at their respective optimized conditions.