Faradaik olmayan elektrokimyasal sensörler ve veri analizleri

This thesis delves into the exploration of non-faradaic electrochemical impedance spectroscopy (EIS) data derived from strong electrolyte solutions using diverse sensing structures. The primary objective is to assess the capability of these data to identify the presence of electrolytes and their respective concentrations within both individual and mixed salt solutions, without using ion-selective elements, reference electrodes, or other intricate setups. Initial experiments involve the manual preparation of solution sets featuring varying concentrations, reaching up to 200 mM. The EIS data is then measured utilizing a Cr-Au electrode-based electrical double-layer (EDL) capacitor. Subsequently, an automated fluidic platform, incorporating bare graphite electrodes, is developed to streamline the entire process, encompassing solution preparation and data collection. The non-faradaic EIS data is measured through voltage sweeps (ZV) within a non-faradaic interval on selected frequency values spanning from 20 Hz to 200 kHz. Implementation of unsupervised and supervised learning methods is carried out for the obtained non-faradaic datasets. Additionally, a novel augmentation model is introduced for the ZV-EIS dataset, making neural network model training possible. The findings demonstrate that supervised learning models yield satisfactory performance results in tasks related to compound classification and concentration detection. In addition, the study delves into the impact of frequency bias and sampling voltage window variations, as well as analysis of electrode selectivity and concentration coverage. The thesis concludes with a discussion on future directions for research in this domain.