27 October 2025 in Nature Nanotechnology
The recent revision of the International System of Units (SI)—which fixed the numerical values of nature’s fundamental constants—has opened new perspectives for practical realizations of SI units. Here we demonstrate an intrinsic resistance standard based on memristive nanoionic cells that operate in air at room temperature and are directly accessible to end users. By driving these devices into the quantum conductance regime and using an electrochemical-polishing-based programming strategy, we achieved quantum conductance levels that can be exploited as intrinsic standard values. An interlaboratory comparison confirmed metrological consistency, with deviations of –3.8% and 0.6% from the agreed SI values for the fundamental quantum of conductance, G 0 , and 2 G 0 , respectively. These results lay the groundwork for the implementation of national metrology institute services on chip and for the development of self-calibrating measurement systems with zero-chain traceability.
15 September 2026 in Research Square
The remarkable adaptability of biological systems stems from their ability to continuously modify their structural organization in response to experience, making learning an inherently embodied and adaptive process. In contrast, artificial intelligence systems implement learning through the adaptation of connection weights within fixed-topology architectures. This hinders their ability to learn and interact with the environment through adaptive structural reconfiguration. Here we report a brain-like self-organizing hardware architecture that processes information, embeds learning, and stores knowledge directly at the matter level through the co-evolution of structural dynamics and functional connectivity. Leveraging electrochemical mechanisms that emulate synaptic morphogenesis and synaptic plasticity, evolving dendritic structures encode and retrieve spatiotemporal information from the environment as dynamically reconfigurable, engram-like memory traces. This enables the in-materia implementation of learning functions directly emerging from material self-organization. Memory-driven adaptation supports associative learning through classical and operant conditioning and solves optimization problems such as path selection for robotic navigation. By bridging information processing, learning, and memory encoding with adaptive materials and physical network dynamics, this work establishes a route toward embodied intelligence at the matter level.