Oxide Thin-Film Transistor and Circuit Modeling Using Artificial Neural Network

The oxide thin-film transistors (TFTs) have developed rapidly in the past years and are entering more and more commercial applications, such as display drivers and dynamic random access memories. In contrast to a vast amount of experimental work, very limited work has been carried out on the modeling, particularly with the newly explored artificial neural network (ANN) approach, which is capable of precise modeling. Here, ANN models are established for both n-type indium-gallium-zinc oxide (IGZO) and p-type tin monoxide (SnO) TFTs. Our ANN models include three layers of neurons in order to balance the accuracy and the complexity. The modeled TFT currents agree with the experimental data over a wide range of more than four orders of magnitude. The relative error of the model in the entire experimental current and voltage ranges is no more than 0.47% and 0.29% for the IGZO and SnO TFTs, respectively. The ability for the model to predict nonmeasured device current also allows for circuit modeling, as evidenced by the agreement between the predicted oxide inverter and nand gate characteristics and the experimental data.