Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain
the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in
Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles
and JavaScript.
Vector–matrix multiplication with monolayer memories
1,024 floating-gate field-effect transistors that have channels made from monolayer molybdenum disulfide can be used to perform vector–matrix multiplication and discrete signal processing. The computer-generated image on the cover shows a section of a wafer used to prepare the processors, where the monolayer memories are arranged in 32 by 32 matrices.
Riki Banerjee, vice president of research and development at Synchron, tells Nature Electronics about the company’s work on brain–computer interfaces and the future of communications.
The integration of high-performance n-type and p-type two-dimensional transistors — which can be fabricated on 300 mm wafers using a die-by-die transfer process — is an important step in the lab-to-fab transition of two-dimensional semiconductors.
A 3D stackable computing-in-memory array that is based on resistive random-access memory could accelerate the implementation of machine learning algorithms.
The monolithic integration of photonic and electronic technology can be used to create miniaturized implantable microsystems capable of high-resolution optical neural control and electrical recording in deep brain regions.
An ultrathin haptic interface can selectively activate different cutaneous receptors in the skin, providing rich haptic sensation information in virtual reality.
For a long time, spin–orbit coupling in bismuthates has been considered to be negligible; however, giant charge-to-spin conversion has now been observed in Ba(Pb,Bi)O3-based heterostructures. These observations provide a path toward investigating the interplay of hidden spin–orbit phenomena and superconductivity.
A polymer-free method for stacking 2D materials has been demonstrated, using a cantilevered transfer support made from metallized silicon nitride. The assembly process, which is compatible with ultrahigh-vacuum operation, results in atomically clean and uniform interfaces.
This Review examines the development of neuromorphic hardware systems based on halide perovskites, considering how devices based on these materials can serve as synapses and neurons, and can be used in neuromorphic computing networks.
This Review examines the development of thin-film transistors for use in displays, sensors, digital circuits and memory, as well as their potential for future application in emerging technologies such as neuromorphic computing.
A spin–orbit torque efficiency of around 2.7 can be achieved in heterostructures based on the bismuthate BaPb1−xBixO3, which can be used to drive magnetization switching at current densities of 4 × 105 A cm−2.
Membranes made of metal-coated silicon nitride can be used to assemble van der Waals heterostructures without a polymer support layer, thus improving cleanliness and allowing assembly at more extreme temperature and vacuum conditions.
An in-memory computing chip for vector–matrix multiplication and discrete signal processing applications can be fabricated using floating-gate field-effect transistors based on monolayer molybdenum disulfide.
Industry compatible solid-state doping of regions between the channel and contacts in carbon nanotube transistors can be used to control device polarity and improve device performance.
Ising- and Potts-model-based simulated annealing can be performed with photon-detector-based neuron circuits and used to solve a range of optimization problems.
A haptic interface that uses thermal, mechanical and electrotactile modes of stimulation to target different receptors in the skin can provide users with diverse haptic sensations, reproducing the tactile information of fine roughness, macro roughness, slipperiness, force and temperature.
An artificial intelligence hardware approach that uses the adaptive reservoir computation of biological neural networks in a brain organoid can perform tasks such as speech recognition and nonlinear equation prediction.