Artificial Synapse Network on Inorganic Proton Conductor for Neuromorphic Systems Applications (1311.0559v1)

Abstract: The basic units in our brain are neurons and each neuron has more than 1000 synapse connections. Synapse is the basic structure for information transfer in an ever-changing manner, and short-term plasticity allows synapses to perform critical computational functions in neural circuits. Therefore the major challenge for the hardware implementation of neuromorphic computation is to develop artificial synapse. Here, in-plane oxide-based artificial synapse network coupled by proton neurotransmitters are self-assembled on glass substrates at room-temperature. A strong lateral modulation is observed due to the proton migration in the nanogranular phosphorus-doped SiO2 electrolyte films. Functional roles of short-term plasticity, including paired-pulse facilitation, dynamic filtering and spatiotemporally correlated signal processing are mimicked. Such in-plane oxide-based protonic/electronic hybrid synapse network proposed here are interesting for building future neuromorphic systems.

• The paper demonstrates that nanopores in phosphorus-doped SiO₂ films facilitate proton migration, achieving an electric-double-layer capacitance of ~3.0 μF/cm at 1 Hz.
• The lateral-coupled IZO transistors achieve a drain current on/off ratio of ~10⁵ and mimic excitatory post-synaptic currents with energy as low as 45 pJ per spike.
• The network successfully models short-term plasticity and spatiotemporal dynamics, paving the way for scalable, efficient neuromorphic processors.

Overview of Inorganic Proton Conductor-Based Artificial Synapse Networks

This paper presents an innovative approach to hardware implementation of neuromorphic computing systems through the development of an in-plane oxide-based artificial synapse network. The focus is on exploiting inorganic proton conductors, particularly phosphorus-doped nanogranular SiO$_2$ films, for their role in facilitating neuromorphic computations.

The work addresses the central challenge of emulating the synapse functionalities, including short-term plasticity, which is less energy-intensive compared to traditional CMOS circuitry. The authors leverage high proton conductivity facilitated by the structure of nanopores in the SiO$_2$ films, which allows for effective lateral modulation through proton migration.

Key Findings and Experimental Results

1. Nanogranular SiO$_2$ Structural Properties:
   • The phosphorus-doped SiO$_2$ films exhibit a nanopore structure that promotes proton migration essential for synapse-like functions.
   • The specific capacitance of the film was recorded at a maximum of ~3.0 μF/cm at 1 Hz, which results from electric-double-layer (EDL) formation at the SiO$_2$/IZO interface.
2. Lateral-Coupled IZO Transistors:
   • The IZO-based devices display transferable electrical properties consistent with field-effect transistor characteristics, achieving a drain current on/off ratio of approximately $10^5$.
   • The artificial synaptic transistors were shown to mimic excitatory post-synaptic current (EPSC) behaviors with energy dissipation per spike as low as 45 pJ, which is notably efficient compared to traditional implementations.
3. Neuromorphic Emulation:
   • The synaptic function emulation involves spike-duration dependent EPSCs and spike-triggered responses that accurately reflect biological processes such as paired-pulse facilitation and dynamic filtering.
   • The system demonstrated its functionality as a high-pass temporal filter, which is advantageous for applications in dynamic, frequency-dependent signal processing.
4. Spatiotemporal Dynamics:
   • The transistors can model spatiotemporally correlated signals, crucial for complex signal processing necessary in neuromorphic systems.
   • The research demonstrated the supralinear EPSC summation, similar to biological systems, enhancing the relevance of this model for real-world applications.

Implications and Future Scope

The introduction of lateral in-plane oxide-based synapse networks on glass substrates signifies an important development in the domain of neuromorphic engineering. The primary implications are both practical, due to the low energy consumption, and theoretical, due to the robust emulation of complex synaptic dynamics such as EPSC and STP (short-term plasticity).

Future developments could aim to reduce device dimensions further while scaling energy efficiency. The potential for utilizing such systems in complex neuromorphic processors is significant, opening up pathways toward creating more efficient, scalable neural networks that mimic human cognitive functions.

Moreover, expanding this work to incorporate a broader range of neurotransmitter-like ionic conduits could enhance the synaptic fidelity of the system. As research progresses, the potential integration of this technology into more compact frameworks like bio-robots and AI-driven devices appears promising.

In conclusion, this paper advances the frontier of neuromorphic system design by bringing artificial synapse networks a step closer to mimicking biological counterparts in both structure and function, providing a robust foundation for future innovations in intelligent computing systems.