ABSTRACT With the continuous development of computer image processing, developing efficient and low‐power computing devices has become a key challenge. Memristors have integrated in‐situ storage and computing capabilities, making them an ideal choice for low‐power image processing computing architectures. However, current memristors are confronted with the dual challenges of poor stability and further reduction in power consumption. Here, we fabricated Sm:HfO 2 thin film ferroelectric memristors, which combined the excellent ferroelectric and dielectric properties of hafnium‐based ferroelectric memristors, and the doping of Sm elements further improved their electrical characteristics. The device demonstrated stable switching characteristics, good hold, and durability for 10 8 cycles, as well as an ultra‐low power consumption of 23.82 fJ. Meanwhile, a full‐hardware image edge detection computing system is constructed by building a 3 × 3 memristor array with devices, and a grayscale image is used for hardware image edge detection calculation. The test result structural similarity index measure (SSIM) is 93.04%, and the software similarity reached 98.54%. This work has reduced power consumption and improved device stability by fabricating devices with hafnium‐based ferroelectric materials doped with Sm elements, providing a solution for high efficiency and high accuracy hafnium‐based ferroelectric brain‐like computing systems.
Li et al. (Thu,) studied this question.