Abstract
3D NAND flash memory constitutes strong competitors for neuromorphic computing due to its high density and mature technology. Neural networks based on 3D NAND flash differential pair have been reported since conventional block erase scheme hinders individual conductance modulation. However, the differential pair scheme halves the efficient synaptic density and erase operation is still inevitable when one of the pair devices reaches its minimum conductance. To overcome these shortcomings, the 1-bit erase scheme was proposed based on gate induced drain leakage (GIDL) effect. But in this scheme, there are risks in unintended erase. In this work, gradient bias is adopted to eliminate the potential risks in the 1-bit erase scheme. Both TCAD simulation and experimental results manifest better inhibition of unintended erase. Based on this improved scheme, a winner-takes-all neural network is constructed, where the efficient synaptic density is doubled and free from the minimum conductance dilemma in previous work based on differential pair scheme.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
