News

Short updates from the semiconductor industry, with context and links to the relevant chapters. RSS

Sort: Newest first · Oldest first

October 2, 2026 Research

Multipole engineering in dielectric metasurfaces to steer optical response

A free webinar will examine how full-wave finite-element simulation and semianalytical multipole decomposition in COMSOL Multiphysics can be used to design dielectric metasurfaces that control transmission, reflection, and absorption. The focus is on silicon and gallium nitride structures, where different multipole mechanisms are arranged to produce anomalous absorption enhancement and q-BIC-related spectral shaping.

Read more → · Optoelectronics & Photonics

October 5, 2026 Research

RAPID keeps DRAM computation in row-parallel format

Researchers from Syracuse University, Friedrich-Alexander-Universität Erlangen-Nürnberg, and TU Dresden have introduced RAPID, a row-parallel arithmetic processing-in-DRAM architecture. The approach adds migration cells for local horizontal movement between neighboring bitlines and inversion cells for in-array logical inversion, allowing DRAM to work on row-parallel, bit-parallel data instead of forcing column-oriented, bit-serial reorganization.

Read more → · How does it work?

October 6, 2026 Research

Monolithic 3D memristor-TFT stack enables electrically programmable reservoir dynamics

Researchers from Seoul National University and Yonsei University have demonstrated a monolithically integrated memristor-thin-film transistor kernel for multimode reservoir computing. By vertically combining Ru/HfO2/TiN memristors with In2O3 channel TFTs, the structure allows the temporal conductance dynamics to be tuned electrically across a broad time scale.

Read more → · How Does It Work?

October 6, 2026 Research

Programmable memory controller aims to speed adoption of new DRAM techniques

Researchers from ETH Zürich, CISPA, and New York University have proposed Terracotta, a framework that pairs custom DRAM command extensions with a programmable memory controller. The approach is designed to reduce the repeated redesign effort that has slowed the deployment of techniques for in-DRAM computation, lower latency, higher parallelism, and improved maintenance and reliability.

Read more → · How does it work?