Carbon Nanotube Based Spike Neuromorphic Devices and Circuits

IEEE Transactions on Circuits and Systems I: Regular Papers, 2000

We present an original method to implement neuro-inspired supervised learning for a synaptic array based on carbon nanotube devices. The device characteristics required to implement on chip learning within a crossbar of carbon nanotube field effect transistors (CNTFETs) as synaptic arrays were experimentally demonstrated and accurately modeled through a specific electrical compact model. We performed electrical simulations of learning for an array of 24 nanotube memory devices corresponding to a 3 input 3 output neural layer that revealed successful learning of separable logic functions within very few epochs, even when a realistic variability of nanotube diameter was taken into account. Such a learning approach opens the way to the use of high-density synaptic arrays as generic logic blocks in configurable circuits.

ACS nano, 2018

This paper presents aligned carbon nanotube (CNT) synaptic transistors for large-scale neuromorphic computing systems. The synaptic behavior of these devices is achieved via charge-trapping effects, commonly observed in carbon-based nanoelectronics. In this work, charge trapping in the high- k dielectric layer of top-gated CNT field-effect transistors (FETs) enables the gradual analog programmability of the CNT channel conductance with a large dynamic range ( i. e., large on/off ratio). Aligned CNT synaptic devices present significant improvements over conventional memristor technologies ( e. g., RRAM), which suffer from abrupt transitions in the conductance modulation and/or a small dynamic range. Here, we demonstrate exceptional uniformity of aligned CNT FET synaptic behavior, as well as significant robustness and nonvolatility via pulsed experiments, establishing their suitability for neural network implementations. Additionally, this technology is based on a wafer-level techniqu...

2009 IEEE/NIH Life Science Systems and Applications Workshop, 2009

The design of a cortical neuron with carbon nanotube circuit elements that performs nonlinear dendritic computations with excitatory and inhibitory post-synaptic potentials is presented. The inhibitory synapse with controllable parameters that implement plasticity is described in detail. The circuit design was simulated using carbon nanotube spice models. Simulations show that the neuron fires as long as the inhibitory post-synaptic potential is weak or absent. Strong inhibitory post-synaptic potentials prevent the neuron from firing.

Abstract—The design of a cortical neuron with carbon,nan- otube,circuit elements,that performs,nonlinear,dendritic com- putations is presented. The circuit design incorporates CNFETs, and,was,simulated,using carbon,nanotube,spice models.

arXiv (Cornell University), 2023

In recent years, neuromorphic computing has gained attention as a promising approach to enhance computing efficiency. Among existing approaches, neurotransistors have emerged as a particularly promising option as they accurately represent neuron structure, integrating the plasticity of synapses along with that of the neuronal membrane. An ambipolar character could offer designers more flexibility in customizing the charge flow to construct circuits of higher complexity. We propose a novel design for an ambipolar neuromorphic transistor, utilizing carbon nanotubes as the semiconducting channel and an ion-doped sol-gel as the polarizable gate dielectric. Due to its tunability and high dielectric constant, the sol-gel effectively modulates the conductivity of nanotubes, leading to efficient and controllable short-term potentiation and depression. Experimental results indicate that the proposed design achieves reliable and tunable synaptic responses with low power consumption. Our findings suggest that the method can potentially provide an efficient solution for realizing more adaptable cognitive computing systems. Impact: The huge amount of data generated by the current society makes it necessary to explore new computing methods with higher efficiency to overcome the bottleneck formed between data storage and processing tasks. Neuromorphic computing aims at emulating the functioning of our brain, which performs both tasks utilizing the same hardware. Here, we propose ambipolar field-effect transistors (FETs) based on carbon nanotubes with a polarizable gate dielectric, capable of providing memory functions reminiscent of neuronal synapses, at both polarities of the device. The ambipolar characteristic doubles the possibilities of previously demonstrated neurotransistors. The short-term and ambipolar behavior of the device can find its place in novel applications in the future. Machine learning-enabled gas sensing is an excellent example, where real-time processing of large amounts of data is beneficial. In addition, interaction with oxidative and reductive gases will result in dual responses due to the ambipolarity of the transistor, along with the possibility of storing the sensing data.

Frontiers in neuroengineering, 2009

Carbon nanotubes might improve neuronal performance by favouring electrical shortcuts. FIGURE 1 | Carbon nanotubes enhance excitability of hippocampal neurons by providing electrical shortcuts. (Left) Dendro-somatic electrical coupling (yellow bolt) arises from summation of dendritic Ca 2+ infl ux triggered by back-propagating action potentials (double arrows). (Right) Carbon nanotubes (CNTs, red lines) can establish intimate contact with neurons causing enhancement of this electrical coupling.

Nature communication, 2018

In contrast to AI hardware, neuromorphic hardware is based on neuroscience, wherein constructing both spiking neurons and their dense and complex networks is essential to obtain intelligent abilities. However, the integration density of present neuromorphic devices is much less than that of human brains. In this report, we present molecular neuromorphic devices, composed of a dynamic and extremely dense network of single-walled carbon nanotubes (SWNTs) complexed with polyoxometalate (POM). We show experimentally that the SWNT/POM network generates spontaneous spikes and noise. We propose electron-cascading models of the network consisting of heterogeneous molecular junctions that yields results in good agreement with the experimental results. Rudimentary learning ability of the network is illustrated by introducing reservoir computing, which utilises spiking dynamics and a certain degree of network complexity. These results indicate the possibility that complex functional networks can be constructed using molecular devices, and contribute to the development of neuromorphic devices.

2011 IEEE/NIH Life Science Systems and Applications Workshop (LiSSA), 2011

A biomimetic carbon nanotube synapse, the portion of the neuron that receives inputs from other neurons, has been fabricated in the laboratory as an analog circuit. The waveforms input to the synapse and output from the synapse resemble biological waveforms in shape and relative amplitudes and durations. This working circuit is an important first step towards the use of nanotechnology for biomimetic neural circuits and neural prosthesis. I.

Biomedical microdevices, 2009

Standard micro-fabrication techniques which were originally developed to fabricate semi-conducting electronic devices were inadvertently found to be adequate for bio-chip fabrication suited for applications such as stimulation and recording from neurons in-vitro as well as in-vivo. However, cell adhesion to conventional micro-chips is poor and chemical treatments are needed to facilitate the interaction between the device surface and the cells. Here we present novel carbon nanotube-based electrode arrays composed of cell-alluring ...

Neuromorphic circuits and systems are mixed analog and digital hardware implementations of biological networks. The Neurons and Synapses form the basic building blocks of these networks analogous to their biological counterparts. The analog VLSI implementation of neuron and synapse circuits with devices working in subthreshold regime ensure compactness and low power consumption in these circuits. The use of advanced technologies of MOSFETs has further improved the area and power efficiencies of the neuromorphic circuits. In this work, we have implemented a Static Excitatory Synapse circuit and an Integrate and Fire Neuron circuit in 180nm technology. The circuits are simulated using Cadence software and verified for the functionality with varying circuit parameters. Further, theStatic Excitatory Synapse circuit is cascaded to an Integrate and Fire Neuron circuit and simulated with adjusted component and parameter values in 180nm MOSFET technology in Cadence software and the Carbon Nanotube Field Effect Technology in HSPICE software as well. The effect of the parameter values on the functionality of the cascaded circuit and the passage of signals is studied,analyzed and compared in the two technologies in terms of power and area efficiencies of the circuits.