Search for: All records

Recent breakthroughs in artificial neural networks (ANNs) have spurred interest in efficient computational paradigms where the energy and time costs for training and inference are reduced. One promising contender for efficient ANN implementation is crossbar arrays of resistive memory elements that emulate the synaptic strength between neurons within the ANN. Organic nonvolatile redox memory has recently been demonstrated as a promising device for neuromorphic computing, offering a continuous range of linearly programmable resistance states and tunable electronic and electrochemical properties, opening a path toward massively parallel and energy efficient ANN implementation. However, one of the key issues with implementations relying on electrochemical gating of organic materials is the state‐retention time and device stability. Here, revealed are the mechanisms leading to state loss and cycling instability in redox‐gated neuromorphic devices: parasitic redox reactions and out‐diffusion of reducing additives. The results of this study are used to design an encapsulation structure which shows an order of magnitude improvement in state retention and cycling stability for poly(3,4‐ethylenedioxythiophene)/polyethyleneimine:poly(styrene sulfonate) devices by tuning the concentration of additives, implementing a solid‐state electrolyte, and encapsulating devices in an inert environment. Finally, a comparison is made between programming range and state retention to optimize device operation.

 

The main goal of the field of neuromorphic computing is to build machines that emulate aspects of the brain in its ability to perform complex tasks in parallel and with great energy efficiency. Thanks to new computing architectures, these machines could revolutionize high-performance computing and find applications to perform local, low-energy computing for sensors and robots. The use of organic and soft materials in neuromorphic computing is appealing in many respects, for instance, because it allows better integration with living matter to seamlessly meld sensing with signal processing, and ultimately, stimulation in a closed-feedback loop. Indeed, not only can the mechanical properties of organic materials match those of tissue, but also, the working mechanisms of these devices involving ions, in addition to electrons, are compatible with human physiology. Another advantage of organic materials is the potential to introduce novel fabrication techniques relying on additive manufacturing amenable to one-of-a-kind form factors. This field is still nascent, therefore many concepts are still being proposed, without a clear winner. Furthermore, the field of application of organic neuromorphics, where bioinspiration and biointegration are extremely appealing, calls for a co-design approach from materials to systems. 

Devices with tunable resistance are highly sought after for neuromorphic computing. Conventional resistive memories, however, suffer from nonlinear and asymmetric resistance tuning and excessive write noise, degrading artificial neural network (ANN) accelerator performance. Emerging electrochemical random-access memories (ECRAMs) display write linearity, which enables substantially faster ANN training by array programing in parallel. However, state-of-the-art ECRAMs have not yet demonstrated stable and efficient operation at temperatures required for packaged electronic devices (~90°C). Here, we show that (semi)conducting polymers combined with ion gel electrolyte films enable solid-state ECRAMs with stable and nearly temperature-independent operation up to 90°C. These ECRAMs show linear resistance tuning over a >2× dynamic range, 20-nanosecond switching, submicrosecond write-read cycling, low noise, and low-voltage (±1 volt) and low-energy (~80 femtojoules per write) operation combined with excellent endurance (>10 9 write-read operations at 90°C). Demonstration of these high-performance ECRAMs is a fundamental step toward their implementation in hardware ANNs. 

Neuromorphic computers could overcome efficiency bottlenecks inherent to conventional computing through parallel programming and readout of artificial neural network weights in a crossbar memory array. However, selective and linear weight updates and <10-nanoampere read currents are required for learning that surpasses conventional computing efficiency. We introduce an ionic floating-gate memory array based on a polymer redox transistor connected to a conductive-bridge memory (CBM). Selective and linear programming of a redox transistor array is executed in parallel by overcoming the bridging threshold voltage of the CBMs. Synaptic weight readout with currents <10 nanoamperes is achieved by diluting the conductive polymer with an insulator to decrease the conductance. The redox transistors endure >1 billion write-read operations and support >1-megahertz write-read frequencies.