Solid-State Ionics Synaptic Transistors for Neuromorphic Computing

Introduction

Neuromorphic computing will revolutionize artificial intelligence for applications such as autonomous driving, smart diagnosis, or natural-language understanding by emulating the operation of efficient biological neural networks. The main challenge in this field is the substitution of conventional transistors for synaptic transistors able to learn in ways similar to a neural synapse, i.e., the development of multistate non-volatile transistors.

Current Challenges

However, currently existing synaptic transistors have been developed using electrolytes that are by nature unstable and difficult to integrate, such as ionic liquids or proton-conducting polymers.

Project Overview

TRANSIONICS will deliver highly stable (non-volatile), silicon-compatible, and scalable solid-state synaptic transistors by exploiting the first-ever room temperature oxide-ionic electrolyte developed in the ERC CoG grant (ULTRASOFC) held by the PI.

Technology Features

TRANSIONICS transistors are able to modulate their channel properties with external stimuli like real neurons by reducing/oxidizing a mixed ionic-electronic conductor unveiled at ULTRASOFC. Additionally, TRANSIONICS is compatible with mainstream microelectronics fabrication technology, which makes it ideal for developing high-density brain-like computer chips.

Project Goals

The goals of the TRANSIONICS project are:

1. To evaluate the technical feasibility for the fabrication of unique all-solid-oxide synaptic transistors with lateral architecture.
2. To assess the silicon compatibility and scalability of the TRANSIONICS transistors.
3. To define an IPR strategy for technology transfer.
4. To build a value proposition for a startup company and to identify customer segments with industrial partners.

Team Composition

To achieve these goals, the PI has joined a project team that combines applied research, technology transfer, and market uptake expertise.