Novel pattern recognition-artificial neural networks applied to synthesis design of La3+-doped BaTiO3 nanosize polycrystals

This paper examines a series of La³⁺-doped BaTiO₃ (LBT) nanosize polycrystalline powders which have been synthesized by sol-gel process. The LBT crystalline size - which is affected by La³⁺ dopant content, drying temperature of quasi-gel and calicination temperature of dried gel - is the most essential factor that influences the humidity sensitivity and fabrication of the sensing element that composes the LBT polycrystals. The sensor, made of 15~25nm powders, displays consistent humidity sensitivity and feasibility of fabrication. Therefore, the samples are divided into two classes according to the crystalline size. Those with the crystalline size from 15nm to 25nm are considered as ideal, while those with the crystalline size less than 15nm and more than 25nm are inconclusive. In order to obtain parameters assuring preparation of the polycrystals with the crystalline size from 15 to 25nm, a combination of a novel pattern recognition (PR) and artificial neural networks (ANN) has been utilized to design the process parameters