Exploring the Potential of Combining Polyoxometalates (POMs) and 2D Materials with Nanogap Electrodes for Neuromorphic Computing.

Neuromorphic computing, inspired by the human brain's remarkable computational capabilities, has emerged as a promising paradigm for advancing artificial intelligence and cognitive computing systems. In contrast to the Von Neumann computing architecture, the human brain relies on neurons and synapses for both storage and computation. Consequently, there has been a growing interest in exploring nanodevices that mimic synapses to achieve highly efficient computing. Among these nanodevices, memristors have gained a lot of attention due to their advantages, such as low power consumption, high integration density and the capability to replicate synaptic plasticity, which align with the requirements of neuromorphic computing. In this work, we explore two nanomaterial classes, namely polyoxometalates (POMs) and two-dimensional (2D) transition metal dichalcogenide (TMD) materials, to facilitate the development of efficient neuromorphic hardware.POMs, a class of nanoclusters composed of metal and oxygen atoms, exhibit tunable multi-redox properties. Their excellent electron-accepting ability, high stability, and photo stimulable redox properties, render them suitable for mimicking synaptic plasticity and being employed in in-memory and neuromorphic computing devices. In this study, we have fabricated resistive switching memory devices based on Keggin-type POMs, bearing various counterions, and the role of these cations on memristive switching characteristics was investigated. The POMs are deposited by spin coating or drop casting on coplanar nanogap metal electrodes. The coplanar electrodes of Au and Al separated by a nanogap of 10 nm are prepared by adhesion lithography. This coplanar nanogap geometry is an ideal platform to accommodate such 1-nm sized molecular clusters. Preliminary results showed a resistive switching behaviour with a low operating bias voltage and high endurance and retention, while control over the reversible redox states can be used in neuromorphic and reservoir computing. Furthermore, our research involved an investigation into POMs, a class of nanoclusters composed of metal and oxygen atoms, exhibit tunable multi-redox properties. Their excellent electron-accepting ability, high stability, and photo stimulable redox properties, render them suitable for mimicking synaptic plasticity and being employed in in-memory and neuromorphic computing devices. In this study, we have fabricated resistive switching memory devices based on Keggin-type POMs, bearing various counterions, and the role of these cations on memristive switching characteristics was investigated. The POMs are deposited by spin coating or drop casting on coplanar nanogap metal electrodes. The coplanar electrodes of Au and Al separated by a nanogap of 10 nm are prepared by adhesion lithography. This coplanar nanogap geometry is an ideal platform to accommodate such 1-nm sized molecular clusters. Preliminary results showed a resistive switching behaviour with a low operating bias voltage and high endurance and retention, while control over the reversible redox states can be used in neuromorphic and reservoir computing. Furthermore, our research involved an investigation into 2D memristors, employing coplanar nanogap electrodes with MoS₂ as the channel material. The MoS₂ is directly grown on to a Si/SiO₂ substrate using atomic layer deposition (ALD) process and transferred onto a coplanar nanogap electrode array. The utilization of MoS₂ in memristors fabricated with nanogap electrodes has demonstrated the potential to reduce switching voltages to a crucial minimum. In conclusion, as we delve into nanomaterials, such as POMs and 2D materials, and their incorporation into artificial neural networks, we can expect significant advancements in neuromorphic computing, bringing us closer to a future, where computational systems mimic the remarkable efficiency and adaptability of the human brain.