Fig. 2: Programmable memtransistors.

a Device-to-device variation in the transfer characteristics, i.e., drain current (\({I}_{{{{{{\rm{DS}}}}}}}\)) versus gate-to-source voltage (\({V}_{{{{{{\rm{GS}}}}}}}\)) for drain-to-source voltage (\({V}_{{{{{{\rm{DS}}}}}}}\)) of 1 \({{{{{\rm{V}}}}}}\) for 250 MoS₂ memtransistors, demonstrating unipolar n-type transport. Distribution of b threshold voltage (\({V}_{{{{{{\rm{TH}}}}}},n}\)) extracted from 250 MoS₂ memtransistors. c Device-to-device variation in the transfer characteristics for 20 V-doped WSe₂ memtransistors, demonstrating unipolar p-type transport. Distribution of d threshold voltage (\({V}_{{{{{{\rm{TH}}}}}},p}\)) extracted from 20 WSe₂ memtransistors. The corresponding means (\(\mu\)) and standard deviations (\(\sigma\)) are denoted in the inset. Memtransistors offers analog and non-volatile memory, where their threshold voltage can be adjusted by applying a programming pulse to the back gate. e Transfer characteristics of a post-programmed MoS₂ memtransistor obtained by applying negative programming voltages (\({V}_{{{{{{\rm{P}}}}}}}\)) of different amplitudes for pulse duration (\(t\)) of 100 \({{\upmu }}{{{{{\rm{s}}}}}}\). f Analog retention characteristics, i.e., post-programmed \({I}_{{{{{{\rm{DS}}}}}}}\) versus time measured at \({V}_{{{{{{\rm{GS}}}}}}}\) = 0 \({{{{{\rm{V}}}}}}\), for five different states. g The impact of programming/erasing on device-to-device variation, demonstrated using different \({V}_{{{{{{\rm{GS}}}}}}}\) sweep ranges. h \({\sigma }_{{V}_{{{{{{\rm{TH}}}}}},n}}\) as a function of \({V}_{{{{{{\rm{P}}}}}}}\) and erasing voltage (\({V}_{{{{{{\rm{E}}}}}}}\)) applied during \({V}_{{{{{{\rm{GS}}}}}}}\) sweeps. An increase in variation is seen for high \({V}_{{{{{{\rm{P}}}}}}}\)/\({V}_{{{{{{\rm{E}}}}}}}\).