Analytical and numerical modeling of electric field in a coplanar devices with neural predictor

The objective of this work is to design, and test a programmable memory device based on an interdigital coplanar electrode structure with two electrodes, which can function as a compact programmable resistor with other artificial neural network hardware. To enable programming of such a device with its capacitive and storage behavior, electric field modeling is carried out, supported by numerical simulation using finite element analysis. To enable geometrical and physical design changes without the need to build additional devices, a neural engine is used that combines both the data from the analytical solution and the numerical solution and predicts the electric field data based on geometrical variations. training of the neural engine was performed using the weight elimination algorithm (WEA). The analytical model for large separation between the conducting electrodes agreed to a great extent with the numerical modeling. The trained neural engine with both analytical and numerical data, showed excellent results in predicting the value of the electric field for different geometrical dimensions. The relationship between the derived analytical solution and the capacitive parallel plate device is also presented. The device was built and tested using polymeric materials.