Deoxyribonucleic Acid Brick Crystal-Based Textile Memristor with a Set Voltage of 0.06 V and a High Switching Ratio as an Artificial Synapse for Neuromorphic Computing

Abstract Memristor synapses based on biomaterials provide the possibility of creating high-performance organic memory devices, which play an important role in next-generation eco-friendly neuromorphic computing. Deoxyribonucleic acid (DNA) has unique electrical conductivity, as well as physical and chemical stability, which is extensively distributed in nature, and it has been used in the resistive switching (RS) layer to prepare memristors. However, the reported DNA-based memristors also require further optimization to improve the performance in terms of set voltage, switching ratio, and energy consumption. Herein, we demonstrate a DNA brick crystal (DNAbc)-based textile memristor, which shows an ultralow set voltage of 0.06 V, a high switching ratio of 107, a low set/reset energy consumption of 0.84/0.50 pJ, as well as good cycling stability and data retention capability. Compared with stranded DNA-based memristors, the DNAbc-based device exhibits superior RS performance, demonstrating the advantage of the DNA brick crystal strategy, which may be related to the ordered packing and arrangement of DNA helices in the DNAbc structure. To further explore its potential for neuromorphic computing applications, the device was demonstrated to emulate typical synaptic functions, including long-term potentiation/depression, paired-pulse facilitation, and the transition from short-term plasticity to long-term plasticity. In particular, a 7 × 7 memristor array was constructed to simulate the signal filtering function based on the spiking-voltage-dependent plasticity characteristic of the device. In image recognition using the CIFAR-10 dataset, the constructed network achieved an accuracy of 92.54%, which remained at 90.96% even when 50% noise was added to the input dataset, with corresponding loss functions of 0.06 and 0.21 under noise-free and 50% noise conditions, respectively. This work lays a foundation for the construction of wearable neuromorphic computing electronics and significantly advances the development of future computing textile systems.