Deep Learning Inverse Design of Phase‐Change Reconfigurable Terahertz Metadevices for Multidimensional Secure Communication

Yisheng Dong, Xieyu Chen, Aarthy Nagarajan, Yi Yang, Zhihao Wang, Rui Yu, Chuang Zheng, Chunmei Ouyang, Xueqian Zhang, Tun Cao, Ranjan Singh, Zhen Tian

ABSTRACT The exponential rise in data exchange and cyber threats in next‐generation 6G networks demands communication systems that are inherently secure at the physical layer. Terahertz (THz) waves combine huge bandwidth with strong directionality, offering a fertile platform for high‐capacity and covert data transfer. Here, we introduce a deep‐learning‐enabled inverse‐design framework that enables the creation of dynamically reconfigurable THz metadevices capable of adaptive, multidimensional encryption. By using a residual neural network trained to directly map target electromagnetic responses to device geometries across continuous material phase transitions, our approach eliminates traditional iterative design bottlenecks and enables rapid, high‐precision generation of versatile meta‐architectures. The resulting devices exhibit multiplexed control over polarization, depth, and phase transitions in Ge 2 Sb 2 Te 5 (GST), enabling eight‐channel encrypted holography with minimal crosstalk and near‐diffraction‐limited fidelity. Furthermore, we realize a reconfigurable diffractive THz neural metadevice that performs universal logic operations under a dual‐key security protocol, requiring both the physical hardware and a digital key sequence for accurate decryption. This combination of intelligent design automation and physical‐layer encryption establishes a new paradigm for secure, high‐capacity, and adaptive THz communication, paving the way for dynamically reconfigurable wireless architectures in 6G and beyond.