Natural Organic Carbohydrate Materials Based Resistive Random Access Memory for Sustainable Neuromorphic Computing Systems

Abstract

Current computing systems are facing two essential challenges: (1) tremendous energy consumption due to the conventional Von Neumann architecture with low energy efficiency; and (2) environmental sustainability by depletion of nonrenewable materials, production of electronic waste, etc. One potential solution to simultaneously address these two issues is by “brain-like” and “green” neuromorphic computing with energy-efficient operation, sustainable material resources, and environmentally friendly disposals. Such neuromorphic computing systems require hardware components not only capable of mimicking human neuron and synapse - the basic building block of biological neural networks, but also made from natural organic materials such as polypeptides (proteins) and polysaccharides (carbohydrates) which are renewable, abundant in nature, and biodegradable. In this paper, we report resistive random access memory (ReRAM) made from encouraging natural organic carbohydrate materials, honey and fructose, for emerging neuromorphic computing systems and neural networks. Honey or fructose resistive films were formed by a low cost solution-based process and sandwiched between bottom and top electrode, a simple metal-insulator-metal structure analogous to a biological synapse with presynaptic neuron (top electrode), postsynaptic neuron (bottom electrode), and synaptic cleft (fructose film). The nonvolatile memory behaviors were demonstrated by modulating the conductance by voltage stimuli applied on the memory device, with excitatory current flow in the device being monitored. Nonvolatile memory characteristics including bipolar resistive switching, retention, endurance cycles, long-term potentiation and depression, etc. were reported. All these results testify that carbohydrate-ReRAM devices are promising for energy-efficient and sustainable neuromorphic systems. Figure 1