An efficient device model for ferroelectric thin-film transistors.

Ferroelectric thin-film transistors (Fe-TFTs) have promising potential for flexible electronics, memory, and neuromorphic computing applications. Here, we report on a physics-based efficient device model for Fe-TFTs that effectively describes memory switching and device I–V characteristics. This model combines a stochastic multi-domain description of FE switching dynamics with a virtual source treatment of TFT device characteristics. It demonstrates that the memory window of Fe-TFTs depends on the amplitude and duration of the applied voltage pulses, thus suggesting quantitative means of programming and control. Additionally, we introduce a machine-learning-enabled method to automatically generate optimal voltage pulses for accurately programming multiple intermediate FE states, which is crucial for multi-bit memory and neuromorphic computing applications. To showcase the model's applications, we simulate a 4 × 4 crossbar array circuit based on Fe-TFTs, highlighting its utility in performing multiply-accumulate computing operations. This small array can achieve a high speed of ∼ 1.28 tera operations per second (OPS) and a power efficiency of ∼ 0.43 W/PetaOPS. The model developed here is valuable for exploring the capabilities of Fe-TFTs in future flexible memory and computing technologies. [ABSTRACT FROM AUTHOR]