Hybrid memory for edge AI; magnetic transistor; laser-powered on-chip gear.
Researchers from CEA-Leti, Université Grenoble Alpes, CEA-List, the French National Centre for Scientific Research (CNRS), the University of Bordeaux, Bordeaux INP, IMS France, Université Paris-Saclay, and the Center for Nanosciences and Nanotechnologies developed a hybrid memory system that combines the traits of ferroelectric capacitors (FeCAP) and memristors into a single, CMOS-compatible memory stack that can support adaptive local training and inference of artificial neural networks.
The unified memory stack is composed of silicon-doped hafnium oxide with a titanium scavenging layer, enabling it to operate as a FeCAP or a memristor depending on how it’s electrically formed. Depending on its state, it can be used for precise digital weight storage (training) and analog weight expression (inference). A digital-to-analog transfer method, requiring no formal DAC, converts hidden weights in FeCAPs into conductance levels in memristors.
This hardware was fabricated and tested on an 18,432-device array using standard 130nm CMOS technology, integrating both memory types and their periphery circuits on a single chip. [1]
Researchers from Massachusetts Institute of Technology (MIT) and University of Chemistry and Technology Prague created a magnetic transistor with built-in memory using a novel magnetic material and an optimization process that reduces the material’s defects.
“In this work, we combine magnetism and semiconductor physics to realize useful spintronic devices,” said Luqiao Liu, an associate professor in EECS at MIT, in a press release. “Now, not only are transistors turning on and off, they are also remembering information. And because we can switch the transistor with greater magnitude, the signal is much stronger so we can read out the information faster, and in a much more reliable way,” Liu says.
To begin with, the researchers replaced the silicon in the surface layer of a transistor with the 2D material chromium sulfur bromide, which acts as a magnetic semiconductor and is stable in air. Due to the material’s structure, it can switch between two magnetic states very cleanly. Changing these magnetic states modifies the material’s electronic properties, enabling low-energy operation.
To create a transistor, the researchers patterned electrodes onto a silicon substrate and then aligned and transferred the 2D material on top. Tape was used to pick up a piece of the material, a few tens of nanometers thick, and place it onto the substrate.
The new transistor can switch or amplify electric current by a factor of 10, a significant improvement over previously developed magnetic transistors. An external magnetic field was used to change the magnetic state of the material. They were also able to control the magnetic states with an electric current, which would enable control over individual transistors. The team plans to further advance this as well as make the method scalable so they can fabricate arrays of transistors. [2]
Researchers from the University of Gothenburg and Chalmers University of Technology built light-powered gears on a micrometer scale for on-chip motors.
Using traditional lithography, the researchers manufactured gears with an optical metamaterial and a diameter of a few tens of micrometers directly on a silicon microchip. By shining a laser on the metamaterial, the gear wheel spins. The intensity of the laser light controls the speed, and it is also possible to change the direction of the gear wheel by changing the polarization of the light.
The second gear from the right has an optical metamaterial that react to laserlight and makes the gear move. All gears are made in silica directly on a chip. Each gear is about 0.016 mm in diameter. (Credit: Gan Wang, University of Gothenburg)
“We have built a gear train in which a light-driven gear sets the entire chain in motion. The gears can also convert rotation into linear motion, perform periodic movements, and control microscopic mirrors to deflect light,” said Gan Wang, a researcher in soft matter physics at the University of Gothenburg, in a statement. “This is a fundamentally new way of thinking about mechanics on a microscale. By replacing bulky couplings with light, we can finally overcome the size barrier.”
Wang is particularly interested in medical applications. “We can use the new micromotors as pumps inside the human body, for example to regulate various flows. I am also looking at how they function as valves that open and close.” [3]
[1] Martemucci, M., Rummens, F., Malot, Y. et al. A ferroelectric-memristor memory for both training and inference. Nat Electron (2025). https://doi.org/10.1038/s41928-025-01454-7
[2] C.-T. Chou, E. Park, J. Ingla-Aynes, et al. Large Magnetoresistance in an Electrically Tunable van der Waals Antiferromagnet. Phys. Rev. Lett. 135, 136702 – Published 22 September, 2025. https://doi.org/10.1103/hpmq-rnh4
[3] G. Wang, M. Rey, A. Ciarlo, et al. Microscopic geared metamachines. Nat Commun 16, 7767 (2025). https://doi.org/10.1038/s41467-025-62869-6
 |
 |
 |
 |
 |  |
 |  |
 |  |
Leave a Reply