Molecular electronics is a field that investigates the electronic properties of single molecules or molecular ensembles and utilizes them as active components in electronic circuits. Molecular electronic devices are promising candidates for future ele...
Molecular electronics is a field that investigates the electronic properties of single molecules or molecular ensembles and utilizes them as active components in electronic circuits. Molecular electronic devices are promising candidates for future electronics because of their intrinsic nanoscale dimensions and the synthetic tunability of molecular functionalities. Furthermore, they enable low-cost fabrication, mechanical flexibility on substrates, and scalable bottom-up self-assembly. Diverse electronic functions can be realized by modulating the energy alignment between the frontier molecular orbitals and the electrode's Fermi level. Although single-molecule junctions provide fundamental insights, large-area molecular junctions based on self-assembled monolayers (SAM) are essential for realizing reliable and reproducible device operation. Among various candidates, redox-active molecules are particularly promising due to their discrete oxidation states that enable reversible and controllable switching behaviors. In this dissertation, I present research on the charge transport characteristics and neuromorphic applications of a three-terminal, ion-gel gated vertical molecular transistor.
First, I investigated the electrical characteristics of the device, where a ferrocene-terminated alkanethiolate (FcC11) self-assembled monolayer is vertically sandwiched between a monolayer graphene source and an Au drain. The ion-gel gated three-terminal FcC11 molecular transistor exhibits a clear gate modulation with p-type-like behavior consistent with highest occupied molecular orbital (HOMO)-mediated transport and shows negligible hysteresis during source–drain sweeps. Compared with control devices made with alkanethiolate (C8), the FcC11 devices display pronounced transfer characteristics, confirming that the gate response originates from the ferrocene moiety. Temperature-dependent two-terminal measurements and cyclic voltammetry support a redox-assisted transport mechanism, showing activation energies (30–100 meV) and reversible Fc/Fc+ redox activity.
Next, I demonstrated artificial synaptic functionalities in the FcC11 molecular transistor. Gate voltage pulses induce an electric double layer at the ion-gel/graphene interface, triggering dynamic postsynaptic-like current responses. The devices exhibited neuro-inspired plasticity, including short-term plasticity such as paired-pulse facilitation and a programmable transition to long-term plasticity upon repeated stimulation. The ferrocene redox moiety was identified as the key enabler of nonvolatile switching behavior, mediating a dynamic, voltage-programmable conductance change via a synergistic mechanism of reversible redox reactions and ion trapping. The devices showed multilevel conductance states with synaptic update characteristics, a crucial attribute for the learning process. As a proof of concept, a neural network simulated with the molecular synaptic transistor achieved ~88% accuracy in MNIST pattern recognition even after a single training epoch. These results establish vertical molecular transistor systems as promising building blocks for molecular-level neuromorphic hardware, with a three-terminal, read/write-decoupled architecture that exhibits synaptic behavior and helps overcome read-disturb of two-terminal memristive schemes.