Answers are grounded in this paper and cite their sources.
Demonstrates enhanced synaptic plasticity control in flash memory, suggesting improved neuromorphic computing capabilities.
Energy-efficient neuromorphic hardware requires precise and reliable control of synaptic weight updates. However, in most charge-trap-based artificial synapses, weight modulation is directly coupled to programming pulse parameters, limiting independent control of plasticity gain and retention. Here, we introduce a drain–gate programming scheme in a MoS2/hBN/Fe2O3 flash memory that enables decoupled control of carrier injection and synaptic weight updates. The device exhibits an on/off ratio of ~107 and maintains clearly distinguishable on and off states over >10 years of extrapolated retention. Beyond conventional gate-driven programming, the proposed drain-gate pulse scheme independently controls the direction and magnitude of conductance updates. As a result, the device exhibits highly linear conductance modulation and achieves 1024 distinguishable conductance states, representing the highest multistate capability reported for flash-memory devices to date. This high analog-state density highlights the potential of the device for high-accuracy in-memory computing and neuromorphic applications. Energy-efficient neuromorphic hardware demands precise synaptic weight control, yet traditional charge-trap synapses struggle with independent plasticity management. Here, the authors introduce a drain–gate programming scheme in MoS₂/hBN/Fe₂O₃ flash memory, achieving 1024 conductance states and exceptional retention, paving the way for advanced in-memory computing and neuromorphic applications.
No takes yet. Share an insight, caveat, or question.
Sung et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: