Electronic Materials, Devices, and Architectures for Energy-Efficient Neuromorphic Hardware

Artificial intelligence is transforming industries ranging from healthcare and autonomous vehicles to smart manufacturing and entertainment. However, its staggering computational demands are fueling a global crisis of energy consumption and carbon emissions. Moreover, the existing von Neumann architecture, with its widening performance gap between logic and memory, poses a serious challenge for advanced technology nodes below 28 nm, as traditional CMOS technology struggles to scale beyond this length. Training advanced computational models such as large language models or deep neural networks requires much faster and lower‑energy (< fJ/bit) access to main memory for frequent weight, error‑gradient, and activation updates. These challenges necessitate new materials, devices, and architectural innovations to enable a scalable, greener, and faster information and communication technology sector for the next generation. This Research Topic consolidates recent advances across a wide range of semiconductor memory technologies for sensing, storage, and computation.This Research Topic highlights recent advances in neuromorphic electronics and addresses several cutting‑edge research challenges by exploring new material compositions, pioneering device structures, benchmarking device performance, investigating novel switching mechanisms, and advancing algorithmic and circuit design strategies. Emphasis will be placed on overcoming key challenges across multiple layers of the technology stack, including device physics, materials science, reliability, and performance metrics such as endurance, retention, synaptic weight update techniques, energy efficiency, scalability, and analog performance, as well as system architecture. This Research Topic also seeks contributions addressing fundamental defect physics, interface chemistry, and computer architecture challenges that advance neuromorphic technologies. By bringing together physicists, materials scientists, and engineers, this Research Topic fosters cross‑disciplinary collaboration to drive memory and computing technologies from fundamental research toward real‑world applications.This Research Topic welcomes high‑quality original experimental or computational studies, review and mini‑review articles summarizing state‑of‑the‑art advances, and perspective articles highlighting future research directions. Cross‑disciplinary contributions addressing the following themes are highly encouraged (but not limited to):- Metal oxides, 2D materials, doped hafnium oxide, hafnium zirconium oxide, and AlScN for emerging device applications.- Volatile and non‑volatile ReRAM, memtransistors, phase‑change memory, and ferroelectric memory devices.- Halide perovskites, chalcogenides, and superlattices for low‑power and sustainable electronics.- Neuromorphic materials, devices, algorithms, and circuits.- Unconventional computing materials and systems, including eco‑state machines, recurrent neural networks, reservoir networks, oscillatory neural networks, spiking neural networks, and support vector machines.- Analog in‑memory sensing and computing architectures, artificial machine vision systems, and retinomorphic devices and systems. Topic editors: Dip Das, Sanjay Kumar, Mrinmoy Dutta Source: https://www.frontiersin.org/research-topics/79246/electronic-materials-devices-and-architectures-for-energy-efficient-neuromorphic-hardware