Optimizing design of the microstructure of sol–gel derived BaTiO3 ceramics by artificial neural networks

Modeling and application of artificial neural network (ANN) technique to the formulation design of BaTiO3-based ceramics were carried out. Based on the homogenous experimental design, the results of BaTiO3-based ceramics were analyzed using a three-layer back propagation (BP) network model. Then the influence of sintering temperatures, holding time, donor additives (La2O3, MnO2, Ce2O3) and sintering aids (Al2O3–SiO2–TiO2 (AST)) on the average grain size (d a), the degree of grain uniformity given by the ratio of the maximal grain size to the average grain size (d max/d a), and the relative density (D r) of doped BaTiO3 ceramics system was investigated. The optimized results and experiment data were expressed and analyzed by intuitive graphics. Based on input data and output data, the sintering behavior of BaTiO3 nano-powder was explained well. Furthermore, the fine and uniform microstructure of sol–gel derived BaTiO3 ceramics with d a ≤ 3 μm, d max/d a ≤ 1.20, and D r ≥ 98% was obtained.