The mercury (Hg) is an element that is naturally present in air, water and soils, however it is one of the most toxic metals, so that severe damage to in the health of living organisms are cause by the exhibition of they whit at small amounts of Hg. The importance its timely detection in the environment is to avoid or lessen the damage to public health. One of the various devices for detecting pollutants are the so-called ion-selective electrodes, which offer advantages over traditional instrumental methods, such as: low prices, short response times, simple construction, among others.

In the present work, the design and development of efficient electrodes for the determination of mercury(II) in aqueous medium is addressed. We should be mentioned that the design and construction of electrodes is not an easy task, since the potential of the electrode is a function of several variables such as temperature, analyte concentration, ionophore concentration, etc., as well as the interactions between them. Therefore, it is necessary to build a broad experimental design to achieve the limit of detection (LOD) and the slope closest to the prediction for the Nernst equation. However, generally into experimental purposes a set of simplified and carried out in the behaviors of the variables are involved with in order to minimize experimental work. As a result of the experimental work, information be able serve as base into developing of new electrodes is generated. Here, three multiobjective optimization models are proposed, in which the concentration of mercury ions, the concentration of ionophore, the concentration of plasticizer and the ion exchanger concentration in the membrane as well as their possible interactions are involved as variables. It should be mentioned that the objectives implied in the models are to minimize the LOD and minimize the difference between the slope and the experimental one.

The results obtained by the models were tested in the laboratory to determine and characterize their behavior, later the model data were compared with the experimental results.