New stacked chip gives wearables local memory, tops 90% accuracy in signal tests
interestingengineering.com Sep 29, 2026

New stacked chip gives wearables local memory, tops 90% accuracy in signal tests

AI-summarised brief · reviewed before publication

Researchers from KAIST, UNIST and POSTECH have created a vertically stacked semiconductor system that integrates carbon‑nanotube transistors with a solid‑state ionogel to provide on‑chip temporal memory. When voltage is applied, mobile ions in the ionogel shift and remain displaced after the voltage is removed, creating a short‑term electrical “memory.” By varying ion concentration and ionogel thickness, the team produced fast‑responding and slow‑responding layers, then stacked them so the upper layer captures recent signals while the lower layer retains earlier inputs. In tests, the stack distinguished all 16 possible patterns of four binary signals and, when used to classify video sequences of moving handwritten digits, achieved over 90 % validation accuracy. The technology could enable tiny wearables to analyze motion and physiological data locally without constant data off‑loading.

💡 Why It Matters

  • · Enabling edge devices to process temporal data on‑chip reduces reliance on external servers, paving the way for truly autonomous, power‑efficient wearables.