Polymer memristors, known for their lightweight and mechanical flexibility, are promising candidates for low-power edge computing. However, the inherent structural inhomogeneity of most polymers often leads to unreliable resistive switching, hindering their production yield and reliability. This study demonstrates a two-dimensional conjugation strategy that achieves a record-high 90% production yield of polymer memristors with miniaturization and low-power potential. By employing coplanar macromolecules with 2D conjugated thiophene derivatives, the π–π stacking and crystallinity of the thin film are enhanced. This results in homogeneous switching across the entire polymer layer, fast responses (32 ns), low device-to-device variation (3.16–8.29%), and scalability to 100 nm with minimal power consumption (10⁻¹⁵ J/bit). The memristor array functions as both an arithmetic-logic element and multiply-accumulate accelerator for neuromorphic computing.

Bin Zhang, Weilin Chen, Jianmin Zeng, Fei Fan, Junwei Gu, Xinhui Chen, Lin Yan, Guangjun Xie, Shuzhi Liu, Qing Yan, Seung Jae Baik, Zhi-Guo Zhang, Weihua Chen, Jie Hou, Mohamed E. El-Khouly, Zhang Zhang, Gang Liu, Yu Chen