Abstract
Neuro-based computing, enabled by memristors and memtransistors, has drawn significant attention, with two-dimensional (2D) molybdenum ditelluride (MoTe2) emerging as a promising candidate for its low phase-change energy barrier and tunable electrical behavior. However, realizing the full potential of MoTe2 for neuromorphic applications has been hindered by the lack of a sufficiently large memory window in pristine devices. In this work, we present an effective method to significantly enhance the analogue switching of lateral MoTe2 devices. This method involves two sequential processes: laser treatment followed by the atomic layer deposition of Al2O3. The synergistic effects of both processes yielded significant performance improvements, with a 70-fold improvement in dynamic range when operating as a memristor and a 20-fold enhancement when functioning as a memtransistor. Consequently, artificial neural network simulations demonstrated a 10-fold enhancement in pattern recognition accuracy, while the dual memristor/memtransistor capability enabled the emulation of both homosynaptic and heterosynaptic plasticity.
Funding Information
  • European Cooperation in Science and Technology (CA20116)
  • Research Council of Finland (13357170)
  • Research Council of Finland (320167)
  • Research Council of Finland (358812)
  • Graphene Flagship (881603)
  • Funda??o para a Ci?ncia e a Tecnologia (2022.15710.UTA)
  • Funda??o para a Ci?ncia e a Tecnologia (PTDC/NAN-MAT/4093/2021)

This publication has 45 references indexed in Scilit: