Load/Path

Geothermal power is easy where hot rock sits near the surface, which is almost nowhere. Drill deep enough and rock is hot everywhere. But mechanical drill bits wear out in hard, hot rock long before they get there.

Quaise Energy's answer is to melt and vaporise the rock with millimetre-wave beams, a method developed at MIT, using the same kind of source fusion researchers use to heat plasma. This week the US Department of Energy picked its Project Obsidian in Oregon for up to $25m. The target is wells above 400 °C. The first three wells should give 25 MW or more, and the aim is 50 MW by 2030.

About 140 GW of US gas turbines burn gas, spin a generator and dump their hot exhaust into the sky. Recovering that heat with a steam system is well understood. But steam kit is big and slow to install, so many plants don't bother.

American Supercritical, a Denver start-up that launched this week with $8m, runs its heat-recovery loop on carbon dioxide instead. Above 31 °C and 73.8 bar, CO₂ becomes "supercritical" and nearly as dense as a liquid. Denser fluid means a smaller turbine and heat exchangers. The company says its unit is a tenth the size of a steam system, needs no turbine changes and adds up to 50% more power from the same fuel. No unit is running yet.

The oceans hold about 4.5 billion tonnes of dissolved uranium, spread very thin. Chemists have made materials that grab it. The unsolved half is mechanical: put huge amounts of that material in the sea, leave it, get it back and do it again, cheaply, in waves.

SuperCritical Materials has partnered with Maha Haji's lab at the University of Michigan. In the university's towing and wave tanks they will test how to pack the material and prototype a system to deploy and recover it. Offshore pilots come later.

Walking robots balance actively, like people. If something goes wrong, a 129 kg machine can topple, and in a warehouse there may be a person in the way.

Agility Robotics designed its new Digit around that risk, IEEE Spectrum reports. When someone comes close, Digit puts down its load and moves into a seated position, so it has almost no height left to fall from. Agility also switched to human-style knees, which make squatting and lifting easier. Digit is 1.8 m tall and lifts 23 kg up to 2.1 m. Agility is aiming for UK and EU sales in 2027.

Solid-state batteries replace the liquid inside a lithium cell with a solid, promising more range and faster charging. China's BYD says it will put them in about 1,000 demonstration cars in 2027. It is targeting about 400 Wh/kg per cell, against 150–190 Wh/kg for its current Blade battery.

The hard part is the factory. The solid electrolyte BYD uses breaks down in moist air, so production needs rooms with a dew point below −40 °C. BYD's own chief scientist calls dendrites and unstable interfaces between layers "the two central unsolved problems". These are company statements from 13–14 September, and they are still being debated this week.

"Anode-free" lithium cells drop one electrode entirely and plate lithium straight onto copper foil. That saves weight and space. But lithium tends to grow into needle-like dendrites that short the cell.

A KAIST team in South Korea borrowed a chip-making patterning technique to cover the foil with tubes about 300 nm wide and 150 nm high. That gives about four times the surface area. A 10 nm layer of MXene, an ultra-thin conductive material, then guides the lithium to spread evenly. Published in Advanced Functional Materials.

Ingestible devices that sense or treat problems in the gut need power. Today that usually means a button battery, which is harmful if it gets stuck.

MIT's Giovanni Traverso and Mehmet Girayhan Say built a cell entirely from food-safe materials: a magnesium anode, a molybdenum trioxide cathode, a gel electrolyte and a wax coating. It gives 1.84 V, lasted about three days in acidic lab conditions, then broke down within a few weeks. It has been tested in pigs and powered a radio tag readable up to 1.5 m away. The wax coating sets how long it survives. Published in Nature Chemical Engineering.

Neuroscientists often need to record brain signals, deliver drugs and shine light into tissue. Each job usually needs its own probe, and every probe causes damage.

Kunyang Sui and Christos Markos at the Technical University of Denmark, working with UCL and others, drew a single fibre that does all three. It is less than half a millimetre across, with eight fluid channels around a core that guides light. Its outlets sit almost 3 mm apart along its length, so one probe can reach different layers of the brain. Tested in live mice only. Published in Advanced Science.

Ovarian cancer is far more survivable when caught early: MIT says five-year survival is above 90% at an early stage and less than half that later. Early tests need living cells from the fallopian tube, but scraping or using enzymes can damage them.

Kripa Varanasi's lab at MIT made a handheld device with two syringes. One makes a vacuum seal on the tissue. The other pushes fluid through a narrow channel, and the flow's shear stress, 1–5 pascals depending on the cell type, lifts cells off intact. So far it has been used on tissue already removed in surgery. Published in Device.