Optogenetics inspired photoelectric memories based on flexible nanogap electrodes (PHOTOMEM 2)

The PHOTOMEM 2 research programme aims to transform the field of neuromorphic computing by creating next-generation devices that integrate light detection, memory, and computation within a single unit. These devices are inspired by the efficiency of the human brain, mimicking biological functions to enable real-time data processing with ultra-low power consumption. This work addresses the growing demand for energy-efficient artificial intelligence (AI) systems, particularly in applications requiring rapid decision-making, such as autonomous vehicles, robotics, and wearable technologies. Building on the successes of the initial PHOTOMEM programme, which developed artificial synapses combining optoelectronic and analogue memory characteristics in perovskite and polyoxometalate-based devices, PHOTOMEM 2 focuses on scaling up these single synapses into neural network arrays and integrating them with photonic circuits. The ultimate goal is to create a complete system for in-memory computing and retinomorphic vision, where devices emulate the retina’s ability to process visual information efficiently. By adopting innovative materials like 2D transition metal dichalcogenides (e.g., MoS2), ferroelectric polymers, and molecular metal oxides, the project aims to optimise performance for advanced applications, including in-sensor computing and Edge AI systems. The Key Objectives of this project are to: Develop nanoscale devices with coplanar nanogap electrodes to achieve high density, minimal energy consumption, and compatibility with photonic sources. Tailor materials to specific applications, exploring novel combinations and heterostructures to improve device performance and scalability. Incorporate light sources directly into the neural network to create compact, optogenetics-inspired computing systems with faster signal processing and reduced latency. The project represents a significant step beyond current state-of-the-art systems by addressing the limitations of traditional heterogeneous machine vision technologies, where data sensing and processing occur separately. Instead, PHOTOMEM 2 proposes a homogeneous approach that co-locates sensing and computation within the same unit, enabling faster processing, reduced latency, and improved energy efficiency. This integration is expected to open new possibilities for smarter machine vision systems with widespread applications. More specifically, the outcomes of this research are expected to impact several key areas, such as: Advanced Technologies: Enabling smarter, more efficient AI systems for autonomous vehicles, robotics, healthcare diagnostics, and environmental monitoring. Sustainable Electronics: Offering eco-friendly electronic devices that consume ultra-low power and support global sustainability goals. Scientific Advancement: Providing the research community with novel methodologies, materials, and device architectures that drive innovation in post-Moore’s Law computing. Economic Growth: Supporting the rapidly growing neuromorphic sensor market that is estimated to expand at a compound annual growth rate of 28% and beyond $550M by 2032 and fostering industrial collaborations for technology translation. Beneficiaries of PHOTOMEM 2 include researchers in materials science, optoelectronics, and AI hardware; industries focusing on Edge AI and sustainable technologies; and society at large, which will benefit from safer, smarter, and more energy-efficient innovations. This work is timely and aligns with the urgent need for scalable, low-power, and sustainable electronics in an increasingly AI-driven world.