Curvature-Variable Image Sensor Array with Mo2TiC2Tx MXene for Dark-field Multiview 3D Reconstruction Application.

Curved image sensors hold great appeal in intelligent artificial vision systems, yet current sensor curving techniques still suffer from certain limitations. Taking inspiration from the cuttability and wrap-like transferability of paper-based electronics in three-dimensional curvy architecture and matched hydrophilic functional materials, we propose an image sensor with variable curvature through a vacuum filtration and folding-pasting process assisted by a thermal release tape mask. This method enables the uniform, patterned, hierarchical, and damage-free integration of Mo2TiC2Tx and silver nanowires onto a paper-based substrate. Both convex and concave image sensors are designed and fabricated: the convex type demonstrates wide-field infrared light source tracking when equipped on a robotic dog, while a concave one enables low-distortion imaging with a single lens. Furthermore, by arranging an array of concave curved image sensors around an object and capturing images from different angles, spatial modeling of the object can be realized even in a dark-field environment, showing great potential in bionic compound eyes and artificial vision systems.