High-performance flexible capacitive sensor based on a carbonyl iron particle/multilayer graphene composite dielectric network for robotic object recognition

Fabricating flexible pressure sensors that combine high sensitivity, a wide linear response range and a low detection limit remains a major challenge in composite materials engineering. Here, we report a flexible capacitive pressure sensor based on a heterogeneous carbonyl iron particle/multilayer graphene (CIP/MLG) dielectric network embedded in polydimethylsiloxane (PDMS). Spherical CIPs and lamellar MLG sheets form a multiscale heterogeneous filler network rich in heterogeneous interfaces and compressible microgaps. This architecture amplifies pressure-induced dielectric modulation and capacitance variation. The optimized sensor achieved a maximum pressure sensitivity of 0.04 kPa⁻¹ across a broad operating range of 0–954 kPa. It also exhibited an ultralow detection limit of 0.318 Pa and maintained stable performance over 6000 loading-unloading cycles. Practical sensing capability was demonstrated through fingertip pulse monitoring, joint-motion tracking and Morse-code-based tactile communication. Moreover, a five-finger sensor array was integrated onto a bionic robotic hand to acquire multichannel tactile signals during grasping. Using these signals, a random forest classifier recognized ten representative objects with 100% accuracy under the present experimental conditions. These results highlight the potential of the heterogeneous dielectric-network design for flexible capacitive pressure sensing in electronic skin and robotic tactile perception.