modeling and simulation of pc-zno tfts using ai/ml techniques

This work develops a machine learning-based model to accurately predict the electrical characteristics of polycrystalline zinc oxide thin-film transistors (pc-zno tfts). a random forest regression model is trained using combined data from multiple drain current versus gate voltage ( ) and drain current versus drain voltage ( ) s1weeps, capturing the complex nonlinear behavior of the device. the model achieves high accuracy, with prediction errors below 1% in most cases, and is validated through comparisons with tcad-simulated i–v characteristics. the full current–voltage (i–v) curves in forward voltage sweeps are predicted well, with high r-squared values of 0.9938 for  and 0.9953 for . this method can replace traditional compact models, which often struggle to capture the variability of pc-zno tfts, by providing a fast, reliable, and scalable modeling approach. moreover, the model can be integrated into circuit simulators such as spice via verilog for device- and circuit-level simulations. this study highlights the potential of machine learning techniques to advance compact modeling and support the development of next-generation electronic displays and flexible devices.