Posts

Showing posts with the label energy efficient

Thermodynamic AI computing mimics natural systems reaching equilibrium to generate images with massive energy efficiency

https://www.livescience.com/technology/computing/thermodynamic-computer-can-mimic-ai-neural-networks-using-orders-of-magnitude-less-energy-to-generate-images "uses orders magnitude less energy, potentially up to 10 billion times less... instead of 0s and 1s leverages natural thermal noise/ physical energy fluctuations... allows set of images to degrade into noise through random interactions, calculating decay process reversal probability, adjusting internal values so reversal as likely as possible, effectively generating image from noise... compared to human brain which operates on only about 20 watts... potential 3rd way of computing distinct from classical/ quantum" Related: Noise-free full-color holography under incoherent ordinary light: a quarter-wave geometric-phase route to practical 3D imaging https://www.eurekalert.org/news-releases/1115487

Optical computer routes light through multi-pass loop of optical elements, similar to how light bounces seemingly infinitely between mirrors

https://www.eurekalert.org/news-releases/1115813 "loop allows light to build up complex relationships naturally over repeated passes, using standard LCD/ LED components... passive or low-power optics drastically reduces heat/ energy... photons do not interact with one another, allowing multiple signals to pass through simultaneously at speed of light... uses abundant/ cheaper materials... edge computing enables smaller, more efficient advanced AI to run on drones, medical sensors, cars without needing cloud connection or draining batteries"

First dissipationless Chern insulator exhibits exotic fractional quantum anomalous Hall effect without an external magnetic field

https://phys.org/news/2026-02-current-loss-newly-fractional-quantum.html "2 layers molybdenum ditelluride twisted at specific angle suffers from energy lost as heat... overcome using: horizontal flux growth increasing charge-carrier mobility >10X, refined fabrication significantly reducing twist-angle disorder between layers... when tuned to 2/3rd filling electronic state, resistance nearly vanished/ electricity flowed along material's edges with essentially no energy loss... unlike standard quantum Hall states, energy gap decreased as magnetic field was applied... low-power quantum devices"  Related: Using light to probe fractional charges in a fractional Chern insulator https://phys.org/news/2026-02-probe-fractional-chern-insulator.html Energy loss triggers quantum thermal Hall-like effect at macroscopic scale https://phys.org/news/2026-02-energy-loss-triggers-quantum-thermal.html Physicists watch light drift in quantized steps for the first time https://phys.org/new...

Osmoses' record-breaking polymer membrane aims to replace industrial chemical separation that uses 10 to 15% of global energy use

https://techxplore.com/news/2026-01-industrial-chemical.html "hydrocarbon ladder 3D polymer membranes with tunable structural backbones filter gases at molecular level without using heat, allowing for small gas molecule separation with record-breaking precision... Smaller footprint/ easier integration into existing facilities (lower upfront costs)... separate: CO2/ methane (landfills, dairy farms), helium/ hydrogen wells, hydrogen/ industrial chemicals... planned pilot projects, and potentially producing hundreds of kg: carbon capture, natural gas sweetening, refrigerant reuse" Related: New perspective charts path to next-generation water and energy membranes https://techxplore.com/news/2026-02-perspective-path-generation-energy-membranes.html

Hafnium(IV) stannide mimics graphene's electronic properties while maintaining a robust 3D structure, avoiding 2D graphene's fragility

https://www.eurekalert.org/news-releases/1114004 "2D-like electron flow in solid 3D network structural stability... honeycomb layers with chiral stacking pattern preserves 2D electronic behavior within 3D volume... hosts Weyl points enabling electrons' energy-efficient movement with extreme ease... create materials where function is dictated by atomic arrangement rather than just physical dimensions... low power computers, spintronics, scalable advanced electronics' manufacturing using materials more durable than layered crystals"

Error-aware Probabilistic Update: Memristor-based hardware train AI models with 6 orders of magnitude better energy efficiency than GPUs

https://techxplore.com/news/2026-01-memristor-method-slashes-ai-energy.html "hardware-algorithm mismatch hinders analog in-memory computing for training deep neural networks: difficult to precisely write to/ backpropagation requires tiny, deterministic weight adjustments but memristor noise/ drift creates update errors 10 to 100X larger than intended adjustment... overcome embracing noise using probabilistic scaling precise updates converted into larger discrete pulses applied with specific probability... skips update if smaller than noise threshold... reduces writes by over 99%"

Graphene-derived, dense-block reduced graphite oxide, highly concentrated 3D printing inks avoid use of performance-degrading additives

https://phys.org/news/2026-01-graphene-derived-material-generation-electronics.html "trade-off between ink concentration/ flow, used polymer binders/ surfactants to prevent clumping but  dead weight ingredients reduce final product's electrical/ thermal conductivity... overcome using 3D blocks of loosely packed layers rather than flat sheets, preventing interlocking into gel stopping irreversible stacking... remain printable at 100–200 (instead of only a few) mg/mL without any binders... high-resolution screen printing on flexible surfaces: circuits/ interdigitated patterns, electrothermal Heaters"

First fully integrated/ tunable standalone magnonic spin-wave chip operates without external magnets for beyond 6G telecom

https://phys.org/news/2026-01-standalone-chip-external-magnets-future.html "100×150 square μm samarium cobalt micromagnets/ magnetic flux concentrators on chip (no bulky external magnets), existing electronics fabrication (silicon-compatible)... magnetic field tunable (11 to 20.5 mT) adjusting distance between magnets/ concentrators, enabling 3 to 8 GHz adjustment, phase shift tuned up to 120° at 6 GHz, stable up to 200°C... future: integrate with Micro-Electro-Mechanical Systems making real-time reconfigurable revolutionizing consumer/ automotive sensors, medical diagnostics"

Wireless smart edge networks allow small, battery-powered drones/ sensors to run AI models using less memory, 10X less energy

https://www.eurekalert.org/news-releases/1112193 "avoids: edge devices being bulky/ power-hungry, latency, battery drain (constant transmission), security concerns... In-Physics Analog Computing performs math using signals from 5G tower or Wi-Fi router broadcasting radio frequency that encode weights of AI model... frequency mixers combine incoming RF with device's local data... uses existing, miniaturized RF components... 96% image classification accuracy... wireless network distributed intelligence, drone swarms (search/ rescue), smart city sensors" Related: How hawks slip through tight gaps: A flight stability trick drones could copy https://techxplore.com/news/2026-03-hawks-tight-gaps-flight-stability.html

Heat pumps will soon be able to store/ distribute heat as needed from storage smaller/ more powerful than a traditional hot water tank

https://techxplore.com/news/2026-01-heat-pumps-will-soon-be.html "when electricity cheap or heat pumps running, heats non-toxic salt hydrate phase change material until melts, absorbing massive thermal energy... when need heat for showers/ morning warmth, material solidifies/ releases stored energy... salt hydrates' higher energy density than water so up to 4X smaller water tank speeds heat transfer, thin recycled aluminum fins last decades (plasma electrolytic oxidation coated), 85% efficient (was 65), >70% reduced charging times, >80% reduced time releasing heat"

Fast/ energy-efficient optical AI computer training using real measurement reinforcement learning instead of digital twin approximation

https://techxplore.com/news/2026-01-free-optical-ai.html "overcomes failure due to difficulty modeling small hardware shifts in hardware, environmental interference of light paths, light propagation too complex for perfect simulation... optical processor learns on-site without need to understand underlying physics, prevents drastic/ unstable changes during training, algorithm reuses data reducing experiments needed to train... efficiently focuses light through random/ unknown diffuser, holography/ aberration correction, image classification, photonic accelerators"

Molecular device bridges gap between molecular electronics/ neuromorphic computing performing functions based on how it is stimulated

https://www.eurekalert.org/news-releases/1111223 "tweaked 17 ruthenium complex molecular geometry/ ion arrangement variations... adaptable: memory, logic gate (for processing), synapse (for learning), digital/ analog... molecules at dense scale unpredictable, overcome focusing on chemical design as computational principle... data storage/ processing within same physical substrate, combines quantum chemistry/ many-body physics... stimulate: electron movement through molecular film, oxidation/ reduction, counterion rearrangement stabilizing electronic states" Related: Molecular chains with bite: Customized carbon nanoribbons cleaner molecular electronics path https://phys.org/news/2026-03-molecular-chains-customized-carbon-nanoribbons.html#google_vignette

High-performance physical reservoir device matches deep learning/generative AI computational power while using significantly less energy

https://techxplore.com/news/2025-12-ai-device-ion-gel-graphene.html "avoids multiply-accumulate operations... ambipolar behavior from interaction between graphene/ ion gel: fast processing from graphene's high electron mobility, gel's ions/ electrons move at different rates creating wide range of time constants (reaction speeds) allowing processing complex data patterns more efficienctly... 1/100th computational load due to fewer required operations... Edge AI running on sensors or smartphones without draining their batteries or requiring massive data centers"

Lowering barriers to explainable AI: Control technique for LLMs reduces hardware and energy requirements by over 90%

https://techxplore.com/news/2025-12-lowering-barriers-ai-technique-llms.html "LangVAE and LangSpace software make LLMs more explainable/ controllable while drastically reducing resource costs... Geometric Mapping: Instead of modifying LLM itself, creates compressed representations of language... treats internal patterns as geometric shapes/ points in space, allowing mathematical model behavior adjustment... lowers barrier to entry to AI development, more scientists can contribute to AI safety/ explainability... makes AI behavior more predictable for mission-critical fields like healthcare"

Spiking neural network AI is fast, lightweight, and low power (energy efficient), with hardware, circuits, and algorithms co-designed together

https://www.eurekalert.org/news-releases/1110024 "mimics neuron Compute-in-Memory: digital logic more accurate but uses more energy than analog... versatile platforms switching between traditional/ neuro-inspired networks based on application... only communicates (sends spike) when electrical charge (membrane potential) reaches threshold... responds to irregular/ occasional events (sparse signals) enabling AI to operate onboard device (like drone), outside large data centers/ cloud... event-Based cameras send data only when pixel change detected" Related: Single coin-battery runs AI, from multi-sensor fusion to high-fidelity voice/ vision, for years https://mailchi.mp/allaboutcircuits/experience-ai-on-a-coin-cell-at-ces-8871931?e=197a4256c6

Combination of 2 similar compounds with different crystal symmetries exhibits complex magnetic patterns known as skyrmion-like spin textures

https://www.eurekalert.org/news-releases/1109423 "structural frustration at interface of 2 compounds: manganese/ cobalt/ germanium, and manganese/ cobalt/ arsenic... translates to magnetic frustration causing atomic spins to twist, creating desired magnetic properties for: high-density hard drives, reduced power consumption for data centers as moving skyrmions requires little energy, materials for fault-tolerant quantum computing" Related: Using magnetic frustration to probe new quantum possibilities https://www.eurekalert.org/news-releases/1113485 Twist-controlled magnetism grows beyond the moiré https://www.eurekalert.org/news-releases/1115996

On-demand reversible electronic switching of quantum materials' topological state achieved in a single crystal, for energy-efficient electronics

https://phys.org/news/2025-12-demand-electronic-topology-crystal.html "topological currents protected by symmetry have significantly less dissipation (energy waste)... addition or subtraction of electrical charge drives reversible topological transition: initial OFF non-conductor insulator: modifying antimony/ tellurium ratio changes electron count/ lattice arrangement, disrupting symmetry, opening energy gap... switching ON metal conductor: depositing thin potassium layer donates electrons, restoring symmetry... then OFF: heating removes potassium, returns to initial insulating state" Related: Surprising effects under ion bombardment: the quantum switch https://www.eurekalert.org/news-releases/1117774

Control of nanostructures and magnetic domain structures of iron-based soft amorphous ribbons for high-frequency power electronics

https://www.eurekalert.org/news-releases/1108300 "energy-efficient: achieves >50% reduction in core loss compared to initial amorphous material... particularly high performance in high-frequency range (up to several tens of kHz) where materials usually suffer from increased energy loss... efficiency of soft magnetic materials in high-frequency transformers/ inductors in: EV drive power supply circuits, AI-focused data centers... future: prototype transformers: test material's integration into actual power conversion circuits"

Single transparent silicon polymer processed, using laser, into either radiative cooling or solar heating material, without requiring electricity

https://www.eurekalert.org/news-releases/1101723 "applying this material to building roofs improves efficiency, reducing annual heating/ cooling energy costs by up to 26.5%... controlling laser's output intensity changes material's optical/ thermal properties: high-power laser transforms surface into white porous structure (SiO2-centered) reflecting sunlight lowering temperature 5.89°C, low-power laser creates black porous structure (SiC-centered) absorbing sunlight heating surface to 58.1°C... also, temperature differences enable thermoelectric generators" Related: Mechanical power by linking Earth’s warmth to space https://www.eurekalert.org/news-releases/1105781 Cement Breakthrough Turns Building Walls into Power Generators https://news.sustainability-directory.com/research/cement-breakthrough-turns-building-walls-into-power-generators/

Room temperature facile anisotropic atomic-level etching of hafnium oxide films without using halogen gases and creating no reaction byproducts

https://www.eurekalert.org/news-releases/1097408 "strong ionic bond between hafnium and oxygen atoms exhibits high dielectric constant, superior thermal stability, wide band gap, optimal for ultrathin semiconductors... low-pressure, high-density plasma generation alternately irradiates with N2/ O2 plasmas... 0.023 to 0.107 nm etch depth/ cycle... formation of Hf-N bonds, nitrogen atoms replace surface oxygen atoms when exposed to N₂ plasma, then exposed to O2 plasma causing bond decomposition... after 20 etching cycles, surface roughness reduced 60%" Related: Scientists use single-step laser ablation to fabricate ultra-uniform structures smaller than 50 nm https://www.eurekalert.org/news-releases/1102178