Pesticide Identification through PCA-ANN Classification of Cyclic Voltammetry using Screen-Printed Electrodes

Abstract

Introduction Pollution caused by the use of pesticides is a growing concern, strongly linked to the expansion of agricultural operations based on fulfilling the food requirement for humans [1]. Therefore, over the previous decade, there has been a huge increase in research within this area, through both the development of new, innovative detection approaches and the improvement of existing techniques.Several analytical methods were used to analyze pesticides, including fluorimetry, capillary electrophoresis, spectrophotometry, mass spectroscopy, and gas or fluid chromatography [2]. Most of these devices present a number of disadvantages; rather bulky, expensive, and require time-consuming procedures. Research interest has increased on the application of electrochemical instruments for pesticide sensing, since electrochemistry takes benefit of its sensitivity and relative selectivity [3]. The response of electrochemical detectors is given straight with electrical signals, without the need to convert to other types of responses. It results in a less complex instrumentation and consequently low measurement cost, but also in portability, allowing the measurements to be carried out in the field. A potentiostat is used for current-voltage measurements, which forms the basis for electro-analysis. While potentiostats play a significant role in current electrochemical research, they are mostly used in the laboratory courses for little application. These devices have a low penetration due to the price, as even the least expensive commercially accessible laboratory potentiostats are selling for more than USD1000 [4]. The main proposal of this study is the development of a convenient, sensitive, selective and low cost electrochemical sensor that could be used for identification of pesticides in a solution. Cyclic Voltammetry Voltammetric techniques are useful tools in analytical sciences, especially in aqueous media research or in the solid-state of electro-active substances. In a Cyclic Voltammetry (CV) system, by sweeping the potential over time and recording the current, a voltammogram curve is achieved that is equal to the current as a function of the potential applied [5]. The application for voltammetric methods for analytical purposes is due to the fact that the current recorded at the electrode of the indicator is directly proportional to the analyte concentration. With regard to pesticide electro-analysis, most of which are organic compounds with electro-active functional groups, CV and pulse voltammetry are often used. CV is quite common among voltammetric techniques and is usually used to explore the electrochemical behavior of new redox systems in the first place, and CV can also achieve micromolar detection limits. Method The initial electrochemical measurements of cyclic voltammetry (CV) and chronoamperometry were performed with an Autolab potentiostat (Metrohm, model PGSTAT302) with Nova software version 1.7, and the data set for each pesticide will be obtained from this device. To voltammetrically detect the pesticides; two kinds of commercial Screen-Printed Carbon Electros (SPCE purchased from Quasense Co., Ltd) and one homemade electrode (using carbon paste Loctite Edag PR406 E&C) are utilized. To meet the objective of developing a low cost pesticide detector, a homemade potentiostat responsible of carrying out the CV is built by following an open source reference [4], and manually soldering the components to the PCB of the referenced potentiostat schematic. The CV measurements are recorded from the developed device using a Python program that linearly sweeps the potentiostat’s reference electrode voltage from -1V to 1V, while capturing the corresponding current at the working electrode. Results and Conclusions Tables 1 shows 5 types and concentrations of pesticides under investigation.Table 1: 5 types and concentrations of pesticides under investigation Pesticide Concentrations (ng mL-1) Chlopyrifos (organophosphte) 100 1200 Dicofol (organochlorine) 100 1200 Cabaryl (carbarmate) 100 1200 Cypermethrin (pyrethrum) 100 1200 Pirimiphos-methyl 100 1200 Similar to spectroscopy methods, patterns discovered in CV information can be considered as a fingerprint of the experimental system being studied, which is unique and coherent [6]. The use of chemometric tools such as Principal Component Analysis (PCA) or Artificial Neural Networks (ANN) can help to identify and process the electrochemical fingerprint shown by the pesticide sample. The proposed approach is based on coupling voltammetric responses obtained from an SPCE on the five kinds of pesticides, using PCA to visualize the dissimilarities between samples, and finally using ANN to classify the obtained results of the PCA score plot in order to identify each pesticide separately.