A memristive nanoparticle/organic hybrid synapstor for neuro-inspired computing (1112.3138v1)

Abstract: A large effort is devoted to the research of new computing paradigms associated to innovative nanotechnologies that should complement and/or propose alternative solutions to the classical Von Neumann/CMOS association. Among various propositions, Spiking Neural Network (SNN) seems a valid candidate. (i) In terms of functions, SNN using relative spike timing for information coding are deemed to be the most effective at taking inspiration from the brain to allow fast and efficient processing of information for complex tasks in recognition or classification. (ii) In terms of technology, SNN may be able to benefit the most from nanodevices, because SNN architectures are intrinsically tolerant to defective devices and performance variability. Here we demonstrate Spike-Timing-Dependent Plasticity (STDP), a basic and primordial learning function in the brain, with a new class of synapstor (synapse-transistor), called Nanoparticle Organic Memory Field Effect Transistor (NOMFET). We show that this learning function is obtained with a simple hybrid material made of the self-assembly of gold nanoparticles and organic semiconductor thin films. Beyond mimicking biological synapses, we also demonstrate how the shape of the applied spikes can tailor the STDP learning function. Moreover, the experiments and modeling show that this synapstor is a memristive device. Finally, these synapstors are successfully coupled with a CMOS platform emulating the pre- and post-synaptic neurons, and a behavioral macro-model is developed on usual device simulator.

• The paper introduces the NOMFET, a memristive device that emulates biological synapses by exhibiting spike-timing-dependent plasticity.
• It employs a combination of gold nanoparticles and organic semiconductor films to achieve dynamic modulation of synaptic weights.
• Integration with CMOS technology and simulation models positions the NOMFET as a promising building block for scalable neuromorphic computing architectures.

Memristive Nanoparticle/Organic Synapstor for Neuro-Inspired Computing

The paper "A Memristive Nanoparticle/Organic Hybrid Synapstor for Neuro-Inspired Computing" investigates the synthesis and functional demonstration of a novel device called the Nanoparticle Organic Memory Field Effect Transistor (NOMFET). This paper aligns with the search for advanced computing paradigms beyond conventional Von Neumann/CMOS architectures by integrating neuromorphic principles.

Overview and Context

The crux of this research is the NOMFET, a memristive device that mimics the behavior of biological synapses. The device demonstrates Spike-Timing-Dependent Plasticity (STDP), a critical learning rule associated with synaptic connections in the brain. STDP is pivotal for unsupervised learning in neural systems and depends on the precise timing between pre- and post-synaptic spikes to modulate synaptic weights.

Neuromorphic computing, which aims to emulate the neural architectures and operational dynamics of the brain, can significantly benefit from nanoscale devices like NOMFETs due to their tolerance to imperfections and variability in performance. This research extends the potential for incorporating NOMFETs with CMOS technology to support large-scale spiking neural networks.

Experimental Findings

The NOMFET architecture comprises a blend of gold nanoparticles (NPs) and organic semiconductor thin films integrated onto a transistor framework. The modulation of the NOMFET's conductive properties—analogous to synaptic weights—occurs through the charge/discharge dynamics of the NP-organic semiconductor system. The NOMFET function hinges on memristance principles, as supported by Chua's theoretical framework, which allows for the device's conductance to be dynamically altered by previous input states and spike sequences.

In demonstrating STDP behavior, experiments highlight that the shape and timing of applied spikes substantially influence synaptic plasticity. The observed STDP curves exhibited variations depending on whether triangular or rectangular pulses were utilized, providing evidence for tunable learning functions driven by memristive characteristics. This plasticity, while less pronounced than some biological counterparts, enhances the scope for synapstors in artificial neural networks.

Integration with CMOS and Simulation

Beyond individual device characterization, the paper integrates NOMFETs with a CMOS platform that emulates the neural spike generation, creating a hybrid system that substantiates the hardware implementation of STDP rules. The researchers also developed a behavioral macro-model for simulating NOMFET behavior within standard electronic device simulators, such as Spectre-Cadence, offering a pathway for incorporating these devices into practical neuromorphic circuits.

Implications and Future Prospects

The potential application of NOMFETs spans various domains where neuromorphic and neural network systems find utility. The successful demonstration of STDP functionality in a simple hybrid device presents a promising step towards realizing energy-efficient and scalable neural computation systems. Future enhancements might include optimizing organic semiconductor materials or nanoparticle capping to reduce operational voltages and align temporal dynamics more closely with biological synapses.

This research advances the integration of organic electronic components into neuromorphic architectures and opens avenues for further exploration of memristive elements in synapse emulation. While technical optimizations are necessary to bring the device performance closer to biological analogs, the demonstrated concepts provide a solid foundation for continued innovation in neuromorphic computing technologies.