BEOL-Compatible Tellurium Films for Optically Stimulated and Mechanically Deformable Artificial Synapses.

Area of Science:

• Materials Science
• Neuroscience
• Computer Engineering

Background:

• The von Neumann bottleneck limits digital computing efficiency, driving research into brain-inspired neuromorphic architectures.
• Existing artificial synapse devices often rely on high-temperature fabrication processes incompatible with advanced manufacturing.

Purpose of the Study:

• To explore centimeter-scale tellurium (Te) films for optoelectronic synaptic devices using a low-temperature, back-end-of-line (BEOL) compatible process.
• To investigate the synaptic plasticity and mechanical deformability of these Te-based devices for artificial neural networks (ANNs).

Main Methods:

• Chemical vapor deposition (CVD) to grow Te films at 200 °C.
• Characterization of optoelectronic and mechanical properties of Te films.
• Demonstration of synaptic plasticity (EPSC, PPF, memory conversion) and ANNs (pattern recognition, associative learning).

Main Results:

• CVD-grown Te films exhibit broadband photo-responsiveness and significant mechanical deformability.
• Optically-stimulated synaptic plasticity, including short-to-long-term memory conversion, is achieved and maintained under deformation.
• High-accuracy (≈90%) pattern recognition and associative learning demonstrated in ANNs using these devices.

Conclusions:

• Low-temperature processable Te films offer a promising route for fabricating deformable, efficient artificial synapse devices.
• These findings pave the way for novel neuromorphic architectures beyond conventional CMOS technology.
• The persistent photoconductivity (PPC) mechanism underlies the observed synaptic operations and memory effects.