A low-thermal-budget MOSFET-based reservoir computing for temporal data classification

• A silicon-based Schottky barrier MOSFET (SB-MOSFET) fabricated under low thermal budget and compatible with back-end-of-line (BEOL) integration is introduced for neuromorphic computing. • The SB-MOSFET exhibits short-term memory characteristics via modulation of Schottky barriers and charge trapping, enabling reservoir computing for temporal data processing. • The reservoir computing system achieves over 98% accuracy in a 5×4 digital classification task after 50 training epochs, and successfully handles waveform classification and prediction tasks. • The high CMOS compatibility of the SB-MOSFET offers significant advantages for large-scale integration, paving the way for energy-efficient reservoir computing hardware.