Proton Conducting Graphene Oxide Coupled Neuron Transistors for Brain-Inspired Cognitive Systems (1510.06115v1)

Abstract: Neuron is the most important building block in our brain, and information processing in individual neuron involves the transformation of input synaptic spike trains into an appropriate output spike train. Hardware implementation of neuron by individual ionic/electronic hybrid device is of great significance for enhancing our understanding of the brain and solving sensory processing and complex recognition tasks. Here, we provide a proof-of-principle artificial neuron based on a proton conducting graphene oxide (GO) coupled oxide-based electric-double-layer (EDL) transistor with multiple driving inputs and one modulatory input terminal. Paired-pulse facilitation, dendritic integration and orientation tuning were successfully emulated. Additionally, neuronal gain control (arithmetic) in the scheme of rate coding is also experimentally demonstrated. Our results provide a new-concept approach for building brain-inspired cognitive systems.

• The paper demonstrates an innovative integration of GO films with IZO transistors to achieve neuron-like behavior, featuring a high current on/off ratio of ~1.9×10^5 and efficient EDL modulation.
• It employs proton conductivity and capacitive effects of GO to mimic key synaptic functions such as paired-pulse facilitation and dendritic integration, underscoring advances in neuromorphic engineering.
• The devices perform neuronal arithmetic and orientation tuning, indicating their promise for scalable, low-power neuromorphic systems that emulate biological cognitive processes.

Proton Conducting Graphene Oxide Coupled Neuron Transistors for Brain-Inspired Cognitive Systems

The paper "Proton Conducting Graphene Oxide Coupled Neuron Transistors for Brain-Inspired Cognitive Systems" presents an exploration into the development of artificial neurons using proton conducting graphene oxide (GO) as a functional material. These neuron-like transistors demonstrate several intriguing computational capabilities akin to natural neuronal systems and highlight the potential for building advanced neuromorphic systems based on unconventional materials.

The implementation of bioinspired, solid-state devices marks significant progress in neuromorphic engineering. The work contributes to this field by introducing an innovative use of GO, characterized by oxygen functional groups endowing it with excellent proton conductivity. Notably, these properties facilitate the emulation of complex neuronal behavior, such as paired-pulse facilitation (PPF), dendritic integration, orientation tuning, and neuronal arithmetic operations through rate coding. These functionalities are central to understanding and mimicking biological information processing mechanisms.

Key Findings and Results

1. Device Structure and Performance: The paper employs GO electrolyte films in combination with indium-zinc-oxide (IZO) transistors to achieve artificial neuron-like behavior. These transistors exhibit a high current on/off ratio of approximately $1.9 \times 10^5$ and a subthreshold swing of 84 mV/decade, indicating efficient electronic modulation via the electric-double-layer (EDL) effect.
2. Proton Conductivity and EDL Modulation: The specific capacitance of the GO film reaches 18 μF/cm² at 1.0 Hz, attributed to the EDL effect. The ability of proton conducting GO films to facilitate efficient EDL modulation is central to low-energy consumption in artificial synaptic devices, making them viable candidates for future neuromorphic applications.
3. Neuronal Emulation: Paired-pulse facilitation is successfully emulated, demonstrating short-term synaptic plasticity similar to biological neurons. When the time interval between consecutive presynaptic spikes is varied, the ratio of the facilitation effect systematically decreases, paralleling natural neuronal responses.
4. Dendritic Integration and Orientation Tuning: The devices demonstrate spatial and temporal summation of synaptic inputs. The orientation tuning experiments show that neuron transistors respond preferentially to specific stimuli orientations, akin to neurons in the primary visual cortex. Such selectivity is vital for processing visual information in brain-like systems.
5. Neuronal Arithmetic and Rate Coding: The incorporation of multiple presynaptic and modulatory inputs allows the neuron transistors to perform computational operations such as multiplication and addition, critical for neuronal gain modulation. Rate coding is investigated by modulating the spike train frequency, aligning with methodologies in neuronal communication studies.

Implications and Future Directions

The research offers a novel conceptual framework for synthetic neuro-inspired architecture by leveraging unique material properties found in graphene oxide films. In practical terms, these neuron transistors could advance the design of more efficient, low-power neural networks capable of complex tasks including pattern recognition and sensory processing. Moreover, the work suggests pathways for future advancements through device miniaturization and integration into 3D architectures, potentially transforming neuromorphic computing technologies.

The discussion on scaling, particularly reducing device dimensions, is pivotal. As device sizes approach nanoscale through advanced photolithography, potential for integration into larger, more complex circuit architectures expands. Additionally, the low-temperature processing methods propose compatibility with flexible substrates, broadening the application spectrum toward adaptable electronic systems.

In conclusion, the adoption of proton conducting GO materials in neuron transistors sheds light on new directions and methodologies in neuromorphic system design. This paper underscores the unison of material science and cognitive architecture emulation, propelling the field toward more realistic and efficient artificial cognitive systems.