Load/Path

Geothermal power works brilliantly where hot rock sits near the surface, which is almost nowhere. Go deep enough and it's hot everywhere, but mechanical drill bits wear out in hard, hot rock long before they get there.

Quaise Energy's answer is to stop grinding and start melting. Its drilling approach uses millimetre-wave energy, the kind produced by fusion-research gyrotrons, to melt and vaporise rock. This week the US Department of Energy picked its Project Obsidian, south of Bend, Oregon, for up to $25m. The target is wells above 752 °F (400 °C) and reservoirs above 572 °F (300 °C). Three test wells should give 25 MW or more, scaling to 50 MW by 2030. A confirmation well is being drilled now.

About 140 GW of US gas turbines run "simple cycle": they burn gas, spin, and dump hot exhaust into the sky. Adding a steam cycle to recover that heat is well understood, but it's big, slow to build and needs re-permitting.

Denver startup American Supercritical, just out of stealth with $8m, runs its second cycle on CO₂ above its critical point (31 °C, 73.8 bar), where it is as dense as a liquid. Denser fluid means smaller turbines and heat exchangers, about a tenth the footprint of steam kit, the company says. The block bolts onto the exhaust with no turbine changes and, by the company's figures, lifts output up to 50% on the same fuel.

The oceans hold roughly 4.5 billion tonnes of dissolved uranium, at a vanishingly low concentration. Chemists have made adsorbent materials that grab it. The unsolved half is mechanical: put huge amounts of that material in the sea, leave it there, get it back and do it again.

SuperCritical Materials has partnered with the University of Michigan, led by mechanical engineer Maha Haji, to tackle exactly that. In the Aaron Friedman Marine Hydrodynamics Laboratory they'll measure how the adsorbent behaves under offshore loads and design prototype deployment and retrieval systems.

Desalination has two nagging design faults. Salt crusts over evaporators and kills their efficiency, and the leftover brine harms marine life when it's pumped back out.

Chunlei Guo's team at the University of Rochester laser-treats metal until it is black and "superwicking", so water races across it. On the sun-heated centre, seawater evaporates. The coffee-ring effect, the reason a spilled coffee dries with a dark edge, carries the salt out to the cooler edges, where it collects as a solid. Tests with real seawater from several oceans ran without efficiency loss and recovered nearly all the salt. Published in Light: Science & Applications and J. Mater. Chem. A.

MIT's flapping-wing microrobot, driven by soft artificial muscles, had the hardware to fly like an insect but a hand-tuned controller that made it fly like a nervous drone. The controller that could fly it well was too heavy to compute in real time.

Kevin Chen's and Jonathan How's groups ran the heavy, robust model-predictive controller offline, then trained a small neural network to imitate it. The student runs fast enough to fly. Result: 447% faster, 255% more acceleration, and ten somersaults in 11 seconds while holding within 4–5 cm of its path. The goal is robots that can search earthquake rubble where quadcopters can't fit.

Most optical switches work by exciting electrons and waiting for them to settle, which caps speed at picoseconds or slower. Harry Atwater's Caltech group went around that. A patterned pump pulse briefly changes silicon's refractive index through the optical Kerr effect, which needs no waiting, and a second beam passing through bends by up to 13°.

The Kerr effect is feeble, so the team patterned amorphous silicon with sub-wavelength nanopillars that trap light long enough to amplify it. Switching speed was limited only by the 74 fs laser pulse. Published in Nature Nanotechnology.

3D chips are full of deep, narrow trenches. The carbon films used to pattern them follow every bump beneath, so fabs have to polish or etch the surface flat before the next step. Each extra step costs money and adds defects.

ASM International's XP8 Vertos Flowable Carbon, launched on 22 September, is deposited by PECVD but behaves like a liquid. It flows into trenches from the bottom up without voids or seams and settles flat on top. ASM says it's already in high-volume production for logic and memory.

"Anode-free" lithium cells drop the graphite anode entirely and plate lithium straight onto copper foil. That saves weight and volume, but lithium tends to grow into needle-like dendrites that short the cell.

A KAIST team borrowed a chip-making technique, secondary sputtering lithography, to pattern copper foil with tube-shaped structures 300 nm wide and 150 nm high, then coated them with 10 nm of MXene. The patterning gives about four times the surface area and guides lithium to plate evenly. Published in Advanced Functional Materials on 1 September.

GM and Cirba Solutions dismantled end-of-life GM packs, shredded them into "black mass" in Ohio, and refined it back to battery grade. The cathode material in the new cells contains 100% recycled nickel, cobalt and manganese. The first cars carrying those cells left Factory ZERO and Spring Hill this month.

Engineers at India's Mepco Schlenk Engineering College replaced half the fine aggregate in M35 concrete with porous zeolite and 1% of the cement with bamboo biochar. Compressive strength rose 7.48% to 38.49 MPa, and the mix absorbed 1.2 g of CO₂ a day in a test chamber. The paper dates from 2024 and is circulating again this week.