Multi-neuron connection using multi-terminal floating-gate memristor for unsupervised learning.

Multi-terminal memristor and memtransistor (MT-MEMs) has successfully performed complex functions of heterosynaptic plasticity in synapse. However, theses MT-MEMs lack the ability to emulate membrane potential of neuron in multiple neuronal connections. Here, we demonstrate multi-neuron connection using a multi-terminal floating-gate memristor (MT-FGMEM). The variable Fermi level (EF) in graphene allows charging and discharging of MT-FGMEM using horizontally distant multiple electrodes. Our MT-FGMEM demonstrates high on/off ratio over 105 at 1000 s retention about ~10,000 times higher than other MT-MEMs. The linear behavior between current (ID) and floating gate potential (VFG) in triode region of MT-FGMEM allows for accurate spike integration at the neuron membrane. The MT-FGMEM fully mimics the temporal and spatial summation of multi-neuron connections based on leaky-integrate-and-fire (LIF) functionality. Our artificial neuron (150 pJ) significantly reduces the energy consumption by 100,000 times compared to conventional neurons based on silicon integrated circuits (11.7 µJ). By integrating neurons and synapses using MT-FGMEMs, a spiking neurosynaptic training and classification of directional lines functioned in visual area one (V1) is successfully emulated based on neuron's LIF and synapse's spike-timing-dependent plasticity (STDP) functions. Simulation of unsupervised learning based on our artificial neuron and synapse achieves a learning accuracy of 83.08% on the unlabeled MNIST handwritten dataset.

CNT-molecule-CNT (1D-0D-1D) van der Waals integration ferroelectric memory with 1-nm2 junction area.

The device's integration of molecular electronics is limited regarding the large-scale fabrication of gap electrodes on a molecular scale. The van der Waals integration (vdWI) of a vertically aligned molecular layer (0D) with 2D or 3D electrodes indicates the possibility of device's integration; however, the active junction area of 0D-2D and 0D-3D vdWIs remains at a microscale size. Here, we introduce the robust fabrication of a vertical 1D-0D-1D vdWI device with the ultra-small junction area of 1 nm2 achieved by cross-stacking top carbon nanotubes (CNTs) on molecularly assembled bottom CNTs. 1D-0D-1D vdWI memories are demonstrated through ferroelectric switching of azobenzene molecules owing to the cis-trans transformation combined with the permanent dipole moment of the end-tail -CF3 group. In this work, our 1D-0D-1D vdWI memory exhibits a retention performance above 2000 s, over 300 cycles with an on/off ratio of approximately 105 and record current density (3.4 × 108 A/cm2), which is 100 times higher than previous study through the smallest junction area achieved in a vdWI. The simple stacking of aligned CNTs (4 × 4) allows integration of memory arrays (16 junctions) with high device operational yield (100%), offering integration guidelines for future molecular electronics.