Easy-to-use procedure to optimise a chromatographic method. Application in the determination of bisphenol-A and phenol in toys by means of liquid chromatography with fluorescence detection

• Published in: Journal of Chromatography A 2018-01, Vol.1534, p.93-100
• Main Authors: Arce, M.M., Sanllorente, S., Ortiz, M.C., Sarabia, L.A.
• Format: Article
• Language: English
• Subjects: Bisphenol-A; D-optimal design; HPLC-FLD; Parallel coordinates; Pareto front; Phenol
• ISSN: 0021-9673
• DOI: 10.1016/j.chroma.2017.12.049

•Multiresponse optimization, using a D-optimal design, reduces the experimental effort.•D-optimal design and parallel coordinates plot easy and useful tools in HPLC-FLD procedures.•A procedure to analyse bisphenol-A and phenol in toys and in serum bags by HPLC-FLD.•An easy code in MATLAB to extract from the whole domain of solutions the Pareto Front.•D-optimal design for a multicriteria optimization with non-continuous factors. Legal limits for phenol and bisphenol-A (BPA) in toys are 15 and 0.1 mg L−1 respectively. The latest studies show that in Europe the content of BPA, which reaches our bodies through different contact routes, in no cases exceed legal limits. But it is true that the effects caused by continued intake of this analyte for a long time and other possible processes that could increase their migration, are still under consideration by the health agencies responsible. A multiresponse optimization using a D-optimal design for simultaneously optimising two experimental factors (temperature and flow) at three levels and one (mobile phase composition) at four levels, in the determination by means of HPLC-FLD is proposed in this work. The D-optimal design allows ones to reduce the experimental effort from 36 to 11 experiments guaranteeing the quality of the estimates. The model fitted is validated and, after the responses are estimated in the whole experimental domain, the experimental conditions that maximize peak areas and minimize retention times for both analytes are chosen by means of a Pareto front. In this way, the sensitivity and the time of the analysis have been improved with this optimization. Decision limit and capability of detection at the limits obtained were 33.9 and 66.1 μg L−1 for phenol and 25.6 and 50.0 for BPA μg L−1 respectively when the probabilities of false negative and false positive were fixed at 0.05. The procedure has been successfully applied to determine phenol and BPA in different samples (toys, clinical serum bags and artificial tears). The simulants HCl 0.07 M and water were used for the analysis of toys. The quantity of phenol found in serum bags and in artificial tears ranged from 15 to 600 μg L−1. No BPA has been found in the objects analysed. In addition, this work incorporates computer programmes which implement the procedure used (COOrdinates parallel plot and Pareto FROnt, COO-FRO) such that it can be used in any other chromatographic optimization.