Capillary electrophoresis determinations of trace concentrations of inorganic ions in large excess of chloride: Soft modelling using artificial neural networks for optimisation of electrolyte composition

Abstract

In this work, using a combination of experimental design (ED) and artificial neural networks (ANN), the composition of a triethanolamine-buffered chromate electrolyte was optimised for determination of sulphate anions in the presence of high chloride excess. The optimal electrolyte, allowing a baseline-resolved separation of sulphate from chloride present in a 1500 multiple excess in less than 170 s, consists of 10 mmol/L CrO(3), 2 mmol/L hexamethonium hydroxide, 10% methanol, and triethanolamine added to adjust the pH to 8.0. The method is suitable to a wide concentration range of chloride (4-1757 mg/L) and sulphate (4-590 mg/L) with linear calibration plots (R(2) = 0.9937-0.9999). Relative standard deviations are less than 2.0% for both anions for migration times and peak areas. The detection limits (hydrodynamic injection of 1 s) were 0.6 mg/L for sulphate and 0.5 mg/L for chloride. The method was successfully applied to determination of sulphate in mineral waters containing a high chloride concentration and to determination of sulphate traces in an anticancer drug injection preparation containing a physiological level of chloride. It was shown that alpha-cyclodextrin as an electrolyte additive has a significant potential for further increasing the separation selectivity for inorganic anions.

Prediction of Liquid Chromatographic Retention Times of Peptides Generated by Protease Digestion of the Escherichia coli Proteome Using Artificial Neural Networks

“…Another prediction method is based on artificial neural networks (ANN). It applies a machine-learning algorithm for solving nonlinear problems ,− and has been used to model the quantitative structure−retention relationships (QSRR) of various analytes in liquid chromatography 10-12 and for predicting migration times in capillary electrophoresis. − A previously reported ANN method for the prediction of the liquid chromatographic retention times of peptides utilized a multiple-layer architecture consisting of 20 input nodes that corresponded with the 20 different amino acids …”

Large-Scale Prediction of Cationic Metabolite Identity and Migration Time in Capillary Electrophoresis Mass Spectrometry Using Artificial Neural Networks

“…[18][19][20][21][22][23][24] This learning algorithm for solving classification or identification problems [25][26][27][28] has been used to model the quantitative structure-retention relationships for gas chromatography 18 and the retention time for reversed-phase high-performance liquid chromatography (RP-HPLC) [19][20][21] or the migration time for capillary electrophoresis. [22][23][24] Previous predictions [19][20][21][22][23][24]26,[28][29][30][31][32] with ANNs used the multiple-layers architecture model with back-propagation learning (Figure 1); the training methodology encountered the problem of generating many local minimums, which resulted in a decrease in the prediction accuracy. To avoid this problem, we used ANN ensemble techniques [33][34][35] to predict the electrophoretic mobility of cations in CE-MS.…”

Determination of sulfate ester content in sulfated oligo‐ and poly‐saccharides by capillary electrophoresis with indirect UV detection

“…Muzikar et al . (2003) reported the determination of trace amount of inorganic anions (e.g. SO 4 2− or NO 3 ‐ ) in the presence of large excess of Cl ‐ using an electrolyte consisting of triethanolamine (TEA)‐buffered chromate with hexamethonium bromide (HMB) as electroosmotic modifier.…”