Neural Networks from Biophysical Applications in Microelectronics Parameters Measurements

Apply of artificial neural networks (ANN) in technology has started at the second half of the twentieth century. This arose as a consequence of extensive research in the field of biophysics carried out in the middle of the twentieth century. These biophysical researches have led to knowledge how a neuron as a basic element of a living being nervous system receives and processes signals. Living organisms use no mathematical model, the neurons processes signals by a simple algorithm, signals are massively parallel processed, the output is the superposition of all parallel processed signals. Artificial neural networks which are formed on these principles are used solving various problems such as pattern classification, clustering, function approximation and optimization. They can replace theoretical models that are hard to be presented analytically with data obtained using these models. In this paper we analyzed electrophysical parameters based on Murata powders and processing consolidated samples technology. Between different electric, dielectric and ferroelectric characteristics we applied our neuro network approach on tgδ (loss factor) and for predicting the grain sizes in the function of T and time (τ) as distribution function between, the experimentally applied sintering temperatures (for example from 1190 to 1370 °C). Here, we continue to use neural networks to predict microelectronic parameters on the thin films around the grains based on the bulk surface measurements. All these extended applications are the quite new tool and method for getting the distributed parameters values at the real microstructure level what is opening new perspectives in microelectronics miniaturization designing.