Search for: All records

Abstract The prevailing von Neumann bottleneck has demanded alternatives capable of more efficiently executing massive data in state‐of‐the‐art digital technologies. Mimicking the human brain's operational principles, various artificial synapse devices have emerged, whose fabrications generally require high‐temperature complementary metal‐oxide‐semiconductor (CMOS) processes. Herein, centimeter‐scale tellurium (Te) films‐based optoelectronic synaptic devices are explored by a back‐end‐of‐line (BEOL) compatible low‐temperature (200 °C) chemical vapor deposition (CVD). The CVD‐grown Te films exhibit prominent semiconducting properties such as broadband photo‐responsiveness accompanying a large degree of mechanical deformability. These characteristics coupled with their scalable manufacturability realize a comprehensive set of optically‐stimulated synaptic plasticity; i.e., excitatory postsynaptic current (EPSC), paired‐pulse facilitation (PPF), and short‐to‐long‐term memory conversion, all of which are well preserved even under severe mechanical deformations. A variety of proof‐of‐concept applications for artificial neural networks (ANNs) are demonstrated employing these deformation‐invariant synaptic features; i.e., high‐accuracy (≈90%) pattern recognition, associative learning, and machine learning‐implemented visual perception. The fundamental mechanism for the synaptic operations is discussed in the context of their persistent photoconductivity (PPC) and its associated memory effect. This study highlights high promise of low‐temperature processable semiconductors for emergent neuromorphic architectures with various form factors beyond the conventional CMOS strategy.