AI-driven ultrafast spectrometer-on-a-chip: A revolution in real-time sensing
Peer-Reviewed Publication
Updates every hour. Last Updated: 17-May-2026 21:15 ET (18-May-2026 01:15 GMT/UTC)
Scientists have created a fingertip‑scale spectrometer‑on‑a‑chip that brings lab‑grade hyperspectral sensing into the near‑infrared range long considered out of reach for silicon. By engineering photon‑trapping textures onto silicon photodiodes and using a neural network to decode their combined signals, the device accurately reconstructs spectra from 640 to 1100 nanometers—well beyond the limits of conventional silicon spectrometers. Despite its tiny 0.4‑mm footprint, the chip delivers ~8‑nm resolution, maintains high accuracy with fewer than 16 detectors, and remains stable even under heavy electronic noise. The team also demonstrated precise hyperspectral imaging of a butterfly dataset, highlighting the technology’s potential for compact biomedical, environmental, and remote‑sensing tools.
For years, researchers have provided contradictory evidence about the “premelting film” of ice, its thickness and whether it even exists. In The Journal of Chemical Physics, Luis MacDowell sought to resolve this contention. He focused on the phase diagram of ice and, using computer simulations, visualized the movements of molecules at the surface. At the triple point, where all three phases are equally stable, a nanometer-thin film appeared. MacDowell proposes that much of the disagreement is due to experiments that unintentionally occur slightly away from equilibrium.
One of the primary challenges with prosthetic hands is the ability to properly tune the appropriate grip based on the object being handled. In Nanotechnology and Precision Engineering, researchers in China have developed an object identification system for prosthetic hands to guide appropriate grip strength decisions in real time. Their system uses an electromyography sensor at the user’s forearm to determine what the user intends to do with the object at hand.
In APL Bioengineering, researchers use a machine learning algorithm to explore whether electroencephalography could be useful for connecting brain signals with limb movements in patients who have lost some or all their limb function. In tests, the researchers equipped patients with EEG monitors and asked them to perform simple movements, using their algorithm to classify the range of possible signals. They found they could detect the difference between attempted movement and no movement but struggled to differentiate between specific signals.